PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247501 (2022) https://doi.org/10.1117/12.2664730
This PDF file contains the front matter associated with SPIE Proceedings Volume AAS100 including the Title Page, Copyright information, Table of Contents, and Conference Committee Page.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247502 (2022) https://doi.org/10.1117/12.2660105
Sensing muscle state is an important method for obtaining human movement intentions. In this paper, air pressure chips were used as the sensitive elements to complete the design of a new type of flexible force sensing system based on the flexible micro airbag structure. The system was used to conduct experiments on 4 subjects to collect the force myography (FMG) signals of five different hand gestures. The gesture classification accuracies of 85.3%, 94.2%, and 99.3% were obtained by using different pattern recognition algorithms of LDA, SVM, and KNN. Preliminary experimental results show that the sensing system designed in this paper can effectively sense muscle states and has great potential for application in human-computer interaction control.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247503 (2022) https://doi.org/10.1117/12.2659294
The usage of X-rays for classifying and diagnosing pneumonia has shown an excellent range of exactness and accuracy. This paper presents a binary-classification model that diagnoses patients who may have pneumonia by inputting their x-ray images and introduces the concepts used to develop that model. Keras and TensorFlow libraries are used in this analysis to produce a convolutional neural network model. The training data set which are used to train the model contains 5216 samples which represent 5216 different patients with either pneumonia x-ray image or normal x-ray image. The testing data set contains 624 samples which show how well the model generalizes on the new dataset. The model produces a classification accuracy of 75% on the testing set.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247504 (2022) https://doi.org/10.1117/12.2659689
In order to solve how the soft field effect in capacitance tomography led to the problem of low quality of image reconstruction, this paper proposes an image reconstruction algorithm based on Improved Multi-scale Residual Network (IMRN), by introducing a multi-scale convolution structure layer information, abundant feature extracting multi-scale empty convolution structure, and then build a change with different expansion rate convolution receptive field. The global feature information is obtained, and the number of network parameters is effectively reduced. The channel attention mechanism is used to weight the extracted features adaptively and filter the redundant information. Finally, the shallow features and the extracted features of each structure are fused to compensate the lost feature information. Simulation results show that compared with LBP algorithm, Landweber iterative algorithm and 1DCNN algorithm, the improved algorithm effectively improves the quality of image reconstruction.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247505 (2022) https://doi.org/10.1117/12.2659854
In this paper, an acceleration measure (AM) for heart rate variability (HRV) is presented. Via the quantitative index AM, the acceleration of change of the normal sinus rhythm can be described effectively. Experiment results demonstrate that via the higher distribution density of experiment results of normal sinus rhythm, the sinus rhythm can be recognized and classified effectively. Hence, AM can be as a valid feature of the sinus rhythm.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Yu Tu, Xiaolong Wang, Zefan Qi, Yinghu Liu, Xiaoxi He
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247506 (2022) https://doi.org/10.1117/12.2659336
Textures are an important element to simulate real scenes. When textures are applied to each scene, the most important step to more realistic simulation of the real world is to solve the sense of repetition caused by texture tiling. Therefore, in order to simplify the process of texture image production and deal with the seams when texture tiling, this paper improves the traditional procedural texture generation algorithm based on the traditional one. First, this paper proposes an image-based method to deal with the seam problem caused by structured texture tiling, and also proposes a new algorithm to solve the texture repetition problem caused by large-scale tiling. The algorithm synthesizes infinite outputs with the same appearance using random texture blocks as inputs, with random textures, such as overlapping rocks, rocks, grass, etc. Experimental results show that the algorithm can obtain high-quality texture output while the image seamless processing steps are substantially reduced.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247507 (2022) https://doi.org/10.1117/12.2659367
Image super-resolution algorithm is a technology that changes the image from low-resolution to high-resolution. It is widely used in various fields, including but not limited to image compression technology, satellite remote sensing imaging, urban traffic monitoring, medical imaging and so on. At present, the method based on deep learning has a very good effect on the super-resolution processing of low-resolution images, but some algorithms still have some problems, such as the loss of detail texture of the reconstructed image, and the large gap between the reconstruction results and the computational resources. To solve the above problems, this paper proposes a network model based on channel attention mechanism. The network model consists of three parts: the first part is a shallow feature extraction block, which is composed of a convolution layer and an activation layer, which is mainly used to extract the low-level features of the input image. The second part is the deep extraction block based on the channel attention mechanism, and the depth separable convolution is added to the local residual block of this part to effectively reduce the huge parameters generated by training. This module mainly extracts the high-level features of the input image. The third part is the reconstruction module, which is used to fuse the original image features extracted from the previous two parts and output the reconstructed image. Finally, the experimental results show that using this method will effectively improve the peak signal-to-noise ratio and structural similarity index of the reconstructed image.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247508 (2022) https://doi.org/10.1117/12.2659977
The ultrasound image is a commonly used imaging modality to treat thyroid nodules due to its rapid imaging speed and the ability for multiple anatomical and soft-tissue visualization. However, considering the variability of the position, shape, size and intensity of nodules, the accurate segmentation of thyroid nodules is challenging. In this paper, we proposed an automatic segmentation network of thyroid nodules by combining fast-RCNN and spatial-channel attentive U-Net. In our model, fast-RCNN is firstly utilized to identify the rough position of nodules. The predicted boxes are then utilized to crop the image patches containing only nodules and then build a new database. Next, a spatial-channel attentive U-Net is designed and trained to realize the nodule segmentation. By comparing the proposed network with other state-of-the-art models, we can gain superior results in discarding similar non-nodule structures and preserving the boundary information of nodules.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247509 (2022) https://doi.org/10.1117/12.2659707
In recent years, driven by the development of deep neural networks, image denoising technology has made great progress. One of the most representative technologies is the deep neural network denoiser based on supervised learning. The denoiser uses noise-clean image pairs as the input and output of a deep neural network, and trains the deep neural network to achieve the denoising goal. However, collecting many high-quality noise-clean image pairs is extremely challenging, which is mainly because (1) it is difficult to collect true and clean images; and (2) changes in motion and lighting make it impossible to align collected image pairs, which limits the widespread application of supervised learning denoising techniques. To solve the above problems, this paper proposes a simple and effective method Self2Align to train a deep neural network denoiser. First, we proposed an efficient deep network model for inter-image alignment. For the collection of original images, we only use noisy images and collected multiple images for different scenes. The trained alignment network was then used to align the original image pairs automatically. Second, the aligned image pairs generated in the first stage were used as training image pairs for the training of the denoising network. In addition, we introduced a new training strategy so that the network can obtain better performance. The proposed Self2Align architecture eliminates the reliance on noise-clean image pairs and reduces the acquisition difficulty of training image pairs in terms of self-supervised training of the network. We explained the feasibility of our proposed method through theoretical analysis and obtained competitive results through experimental verification.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750A (2022) https://doi.org/10.1117/12.2659359
As global environmental problems are becoming more and more serious, satellite image segmentation and automatic monitoring of forests and water bodies are becoming more and more important in order to prevent natural disasters and monitor natural conditions, including monitoring deforestation and areas affected by floods. In this paper, the traditional U-Net model is improved and a FW-U-Net model suitable for forest water segmentation is proposed. After verification, the segmentation verification scores of the forest coverage area and water area performance test set are 84.37% (83.65%) and 87.92% (85.42%) respectively, which has high accuracy and reference value for forest and water satellite image segmentation.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Jinlin Liu Sr., Bo Xing Sr., Shilin Sun Sr., Zongtang Zhang Sr.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750B (2022) https://doi.org/10.1117/12.2659360
The extraction path of the weak line spectrum in the LOFAR spectrogram under low signal-to-noise ratio conditions is typically about NP-hard combinatorial optimization, which is crucial to determine the motion characteristics of the target. Classical particle swarm optimization (PSO) algorithms have continuous optimization problems, and in this paper, a set-based algorithm for line spectrum extraction of improved discrete particle swarm (S-PSO-LSE) is proposed. The algorithm treats the discrete search space as a point set defined by each path node, while updating the definition of the operator on the set, and a new fitness function is proposed as a line spectrum quality standard. This increases the convergence accuracy for searching the line spectrum, improves the convergence rate, and provides good global search capability. The effectiveness and accuracy of the algorithm for weak line spectra are verified by simulation and sea trial data.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750C (2022) https://doi.org/10.1117/12.2659857
Diabetic retinopathy is a serious ocular complication caused by diabetes, has now become the main reason for the workforce blinding, using digital fundus image for diabetic retinopathy screening regularly is the key to preventing blindness. However, in the process of image acquisition, generation and transmission, it is susceptible to low light conditions and Gaussian noise, and it is difficult to detect small lesions and blood vessels in fundus images, which greatly reduces the accuracy of computer-aided diagnosis. Therefore, this paper proposes an image enhancement algorithm for digital fundus images, using adaptive clipping based on improved Canny edge detection to perform square clipping around fundus region. The improved CLAHE technique was used for low light enhancement to highlight the details of the lesions in the image. Aiming at the additive white Gaussian noise caused by medical image digitalize process, the fundus image denoising algorithm based on self-supervised EM-GMM is used to suppress the influence of noise by imposing sparsity constraint on covariance eigenvalues. The experiment shows that our methods achieve good performance in DDR dataset.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750D (2022) https://doi.org/10.1117/12.2659722
With the development of Internet of Things (IoT) and artificial intelligence, people’s demand for location-based services is increasingly urgent. Aim at recognize human lower limb posture accurately, posture recognition based on image information and sensor information such as magnetometer, accelerometer, and rate gyro, MARG sensors, has been the research hot spot. However, the image methods require higher computing resources and will be affected by environmental factors such as fiber optics, shading, etc; the sensor methods have low accuracy with traditional arithmetic. Kalman filter and Attitude and Heading Reference System (AHRS) is used to extract accurate posture information from sensors’ raw data. The corresponding testing platform is set up based on MPU9250, which is a typical and low-cost motion tracking integrating circuit (IC) of MARG sensors and a ZigBee wireless communication module called E18-MS1-PCB.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750E (2022) https://doi.org/10.1117/12.2659376
Multimodal information fusion can extract the important feature information of each modality and realize the information complementation between the modalities, which is a research direction of considerable interest at present. Extracting texture information from visible images and extracting typical features from infrared images are the main roles of dual-modal image fusion. To address the problems of information loss and low clarity of current deep learning-based fusion methods, this paper proposes a method to fuse bimodal images by embedding a network with attention mechanism. Due to the problems of low resolution and noise in infrared images, this method specifically uses encoder structures with different depths for infrared and visible images to extract shallow and deep features of each image, and then passes them through the attention-based fusion network to obtain the fused feature maps. A one-dimensional convolutional model based on a local cross-channel interaction mechanism without dimensionality reduction is used to construct the channel attention module in the fusion network, thus reducing the network complexity and improving the overall performance. The final fused image in this method relies on the decoder structure to generate. Experiments show that the method in this paper can fuse bimodal information better, and has some visual improvement under subjective evaluation compared with the comparison algorithm, and has 7.82%, 9.46%, 19.14% and 13.85% improvement in various objective evaluation indexes compared with DeepFuse, DenseFuse, FusionGAN and GANMcC respectively.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750F (2022) https://doi.org/10.1117/12.2659313
Image restoration is a hot issue in the field of image processing. Traditional algorithms approach the original function by a regular function or remove redundant noise by similarity. This process is complicated and cumbersome. In this paper, the low rank approximate matrix of the image matrix is equivalent to the product of two smaller matrices. At the same time, the first-order and second-order statistical information of the image matrix is effectively maintained by using the matrix Frobenius norm and matrix kernel normal. Secondly, the alternating direction multiplier method is utilized to solve the model. Finally, experimental results test the effectiveness of the proposed algorithm.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750G (2022) https://doi.org/10.1117/12.2659356
Aiming at the problems of low accuracy and complex algorithm design for defect detection and dimensional measurement of metal parts, this paper designs a duplex inspection system using machine vision technology, which is connected to a multi-channel acquisition board and can achieve high efficiency in image acquisition and processing. The HALCON software is used to pre-process the images with algorithms such as threshold segmentation, morphological processing and image enhancement, and the MLP classifier is used to classify defects in the defect detection. The test results show that the system achieves high accuracy and high efficiency quality inspection, which brings some value to promote industrial automation and intelligent development.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Song Xu, Changjie Xu, Lihong Tong, Qiwei Wan, Haibin Ding
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750H (2022) https://doi.org/10.1117/12.2659640
As the basic part of the field of image recognition, image edge contour extraction technology is facing more and more challenges with the deepening of image recognition technology in daily life. Based on multi-operator fusion ecological expansion method, an image edge extraction algorithm was proposed in this paper. The hybrid algorithm can eliminate the influence of some unfavorable conditions such as background noise, illumination reflective noise and extract the edge contours of single or multiple objects more accurately. According to the geometric characteristics of lines and arcs, we summarized and put forward a general method for judging whether a point sequence is a straight line or an arc, and a problem is solved based on this method, which proves the superiority of this method in the classification of contour types. Combined the existing research results and practical experience, we summarized a set of sub-pixel interpolation methods with high precision, high efficiency, and high feasibility.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Sen Wang, Simin Yang, Yuheng Wu, Xueyan Zhou, Zheng Li
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750I (2022) https://doi.org/10.1117/12.2660685
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750J (2022) https://doi.org/10.1117/12.2659497
Globally, pneumonia is the leading cause of death for young people and children. An x-ray of the chest is usually used to diagnose pneumonia by a trained specialist. However, the process is tedious and can result in disagreements among radiologists. It is possible to improve diagnostic accuracy through the use of computer-aided diagnostic systems. In this work, the ResNet model was selected to work as the COVID-19 and pneumonia detector based on x-ray image. Several experiments are conducted on to achieve an optimal result.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750K (2022) https://doi.org/10.1117/12.2659346
COVID-19 has become an epidemic that has a strong influence on people all over the world. Many methods, like nucleic acid test, are proposed to discover the infected as early as possible in order to prevent further infection. However, as all-involvement nucleic acid test becomes more and more frequent in China, the quality of testing cannot be guaranteed, and the efficiency is quite low. Moreover, there is still a period of time from one who is infected takes a test to he or she is found infected and separated from others, which means that more people are prone to being infected during this period. Furthermore, false positive and false negative problems are worth attentive consideration, and the consequences are inestimable. [1] show that, in a population with a low prevalence of COVID-19 cases, even a highly sensitive and specific test returns many false-positive results.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750L (2022) https://doi.org/10.1117/12.2659347
Transparency is an important parameter to describe the optical properties of lake water. Based on the field measured data of April 19, 2018 and September 8, 2019, and GF-5 hyperspectral satellite images, Random Forest and BPNN methods were used to retrieve the transparency of water bodies. The study shows that the Random Forest algorithm performs well in the waters of Qiandao Lake, the R2 between the inversion value and the measured value is 0.8651, and the MAPE is 0.16m. Based on the GF-5 hyperspectral satellite image on September 8, 2019, the spatial distribution characteristics of water transparency in Qiandao Lake were obtained by using Random Forest algorithm. The results show that the overall transparency of Qiandao Lake is higher (1.3~6.8m), and the water transparency in the middle of the lake is higher than that in the northwest and southwest tributaries. According to the analysis of the measured water transparency and the environmental factors of synchronous measurement, the water transparency of Qiandao Lake is affected by many environmental factors, among which the correlation with the concentration of suspended solids is strong, and the correlation coefficient is 0.831. The correlation with total phosphorus concentration, water surface temperature, ph value, wind speed and other environmental factors is weak.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750M (2022) https://doi.org/10.1117/12.2659344
Aiming at the problems that hidden defects inside objects are difficult to be visually recognized and the defect samples obtained from inspection are few, a small-sample learning detection model using ultrasonic flaw detection to extend machine vision is proposed. The model introduces an attention mechanism into the deep nearest neighbor network to adjust the image features, so that the model pays more attention to the useful defect area features, increases the amount of defect-related information, and makes full use of key defect features to detect image targets. Experiments show that the proposed method has the best performance compared with the baseline model on the self-made hidden defect dataset, and the average correct rate is up to 83.85% under 10-shot; the model is tested with noisy images, and the results show that the model detection under noisy conditions has an accuracy rate of about 76%, and it has a certain anti-noise interference ability.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750N (2022) https://doi.org/10.1117/12.2660297
Coal and gas disasters occur frequently, which not only cause casualties, but also bring economic losses. The prediction of coal and gas outburst has important research significance. In this paper, a security combination prediction model for building projects based on PCA-PSO-SVM is proposed. Principal component analysis was used for dimension reduction to remove the principal component with low contribution rate, PSO was used to avoid the blindness of selecting parameters of SVM manually. The average prediction accuracy of this model is 93.85%. Compared with the traditional method, the prediction accuracy and the calculation speed is faster.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Signal Processing and Information Identification and Detection
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750O (2022) https://doi.org/10.1117/12.2659628
The illegal flight of drones have occurred frequently, which pose a great threat to the public security, and there is an urgent demand to develop related detection and identification technology to prevent such illegal behavior. In this paper, we analyze the principle and influencing factors of drone RF recognition, and propose a drone RF recognition method based on the deep attention mechanism. In our work, a fully convolutional network with the encoder-decoder architecture is adopted, and the residual network is used as the backbone network for feature extraction. Moreover, an RF channel attention aggregation (FCA) module is specifically designed for recognition. Our model was trained and verified on a public dataset (DroneRF Dataset), and achieved 99.895% accuracy for drone detection, more than 98.61% accuracy for drone recognition, and more than 99.33% accuracy for drone working mode recognition. On the self-test dataset, the performance of the RF recognition method proposed in this paper was further verified using the transfer learning approach.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750P (2022) https://doi.org/10.1117/12.2659699
Person re-identification is widely considered as a sub problem of image retrieval. It is a technology that uses computer vision technology to determine whether there is a specific pedestrian in an image or video, given a monitored pedestrian image, the pedestrian image under the cross device is retrieved. Many researches have been made to study how to match one people from two picture, but how to locating proper body parts has become a problem which could lead to misalign and wrongly divided sample. To solve this problem, we introduce a pose-auxiliary network to extract the position feature of pedestrian. We use a post estimation network to estimate the person’s body location, then we decide whether it’s a body part or a static obstacle by using a separated trained classifier. Then we compare these features from two picture and calculate their loss. Meanwhile, we use a graph-matching strategy to compare two pedestrians pose structure, which is also used into jugging the match problem. By using a multi-verification strategy, our method has some improvement than some previous study using Market-1501 and DukeMTMC-Reid datasets, but still remains room for improvement.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750Q (2022) https://doi.org/10.1117/12.2659623
Interrupt process and trigger is to some extent the main way computer system interact with external device or system. Device with operating system could complete done this by system core and drivers, but soc or embed system running program directly need to handle the things themself. For the fact that rapid response and random interrupt occasion, it could not ensure completely the program for function data unmodified between before and after interrupt. Thus interrupt spot protection as operating system interrupt function but running directly in soc system using only necessary program is important for the some environment. This article propose a way to realize interrupt spot environment protection in soc or embed system, divide memory to two subs to realize interrupt and program data both integrate among interrupt process. Finally test by experiment and conclude result interrupt protection use 8% extra processor time and realize data correctness as expected.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750R (2022) https://doi.org/10.1117/12.2659783
With the help of neuroscience research, the neural network model based on the pulse generation time can be accurately established, which is called Spiking Neural Network (SNN). This new type of neural network uses spike coding, which allows for more information and greater computing power to be obtained by obtaining the precise timing of the spikes. The application of SNN in music melody generation has a good application prospect. In this study, the Cell of Long Short-Term Memory (LSTM) in the generative model was replaced by Leaky Integrate-and-Fire (LIF), and two different variants of LIF were proposed based on Cell. Although the LIF variant achieved better results than the original LIF, it was not better than Cell in comprehensive comparison, and replacing Cell with LIF in the generative model could not improve the quality of the generated melody. The simple strategy used in the experiments to solve the supervised training problem of LIF was the main reason for such results. For SNN to be widely used, it is urgent to solve its supervised training problem.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750S (2022) https://doi.org/10.1117/12.2660261
At this stage, the PM2.5 concentration prediction algorithm ignores the influence of other air pollution factors, and has not realized the time-dependent integration with the influence of other environmental pollutants. In this regard, the PCA-EDWaveNet-LSTM algorithm considering other air pollution characteristics is proposed. The algorithm proposes to consider other air pollution factors, combine the influence of other air pollution factors with the times dependence on PM2.5 particle concentration, and establish a PCA-EDWaveNet-LSTM algorithm based on air pollution characteristics. In the empirical analysis of PM2.5 historical concentration prediction in Xi’an, the algorithm is compared with RF_Regression algorithm, SVM algorithm, and LSTM neural network. The results show that the prediction performance of this algorithm is better than various traditional prediction algorithms in PM2.5 concentration prediction.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750T (2022) https://doi.org/10.1117/12.2659354
To manage the credit risk in supply chain financing (SCF) more effectively and maintain the healthy operation of the supply chain, this paper proposes a prediction model of credit risk in SCF based on IG-GA-SVM. Firstly, the information gain (IG) is used to extract the feature indexes of the original index system, and then the genetic algorithm (GA) is used to optimize the parameters of the support vector machine (SVM). Finally, the feature indexes selected by the information gain are input into the GA-SVM model for training, and the final prediction model is established. The test results show that the performance of the prediction model based on IG-GA-SVM is better than that of BP, SVM, and GA-SVM models. It has good generalization ability and can provide strong decision support for financial institutions to provide financing for small and medium-sized enterprises (SMEs).
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750U (2022) https://doi.org/10.1117/12.2659625
Sentiment analysis has gradually become an important content of natural language processing (NLP), and plays an increasingly important role in the fields of system recommendation, user emotion information acquisition, and public opinion reference for governments and enterprises. In the period of comprehensive well-off, the leisure consciousness of Chinese residents has been significantly improved, the income growth has brought about the release of leisure consumption potential, and the time guarantee for leisure has been further enhanced. As the urban public cultural space, the modern Science and Technology Museum bears the diversified spatial functions of knowledge production, cultural empowerment and public welfare. An appropriate range of commercial service supply is an important part of the public policy supply of the Science and Technology Museum. It is very important to understand the emotional tendency of the public for the commercial service of the Science and Technology Museum. Roberta adds a dynamic mask mechanism on the basis of the model Bert, taking a larger amount of pre training data and a larger batch size. This paper introduces a multi-channel mask mechanism on the basis of the Roberta model, and increases the mask ratio, so that the model can learn more levels of emotional information, and the effect on text sentiment analysis is better. Therefore, taking Shanghai Science and Technology Museum as an example, the Roberta model is used to extract and interpret the public perception data of the public comment network, and study the value perception and emotional tendency of the public to the commercial services of the Science and Technology Museum, so as to better guide the Science and Technology Museum to improve the service quality level.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750V (2022) https://doi.org/10.1117/12.2660074
According to the characteristics of Tibetan named entity itself, this paper analyzes the text characteristics and recognition difficulties of named entity, and puts forward BiLSTM-CRF model. The combination of the two models not only makes use of the advantages of bidirectional LSTM model to save context information, but also reduces the influence of CRF layer from sentence level to consider before and after annotation, so as to achieve the effect of learning from each other's strengths and complementing each other's weaknesses, which is more effective to solve the problem of Tibetan named entity recognition. The experimental results show that the proposed recognition method has good performance, and the f-value can reach 88.7%.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750W (2022) https://doi.org/10.1117/12.2659584
COVID-19 has now become one of the most severe and acute diseases worldwide. Novel Coronavirus transmission is characterized by its high speed and large social population base, making novel Coronavirus detection very difficult. Therefore, automatic detection systems should be implemented as an option for rapid diagnosis. Automated disease detection frameworks help physicians diagnose diseases with accurate, consistent, and rapid results, and reduce ethics. In this paper, we propose a deep learning method based on long-term Memory (LSTM) for automatic diagnosis of COVID-19 in combination with the existing prediction model SEIR.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750X (2022) https://doi.org/10.1117/12.2660122
Ridge regression is an important algorithm in machine learning, which has a wide of practical applications such as recommendation systems. Therefore, ridge regression outsourcing scheme has been widely studied in recent years. However, it would be difficult to outsource ridge regression to an untrusted cloud server without giving away information. In this work, we propose a new secure and effective outsourcing ridge regression scheme. We use a series of disguise techniques with permutation matrices and unimodular matrices for encryption. Through theoretical and experimental analysis, our protocol has two advantages: 1) The protocol guarantees against malicious cloud server attack and the client can verify the results returned by the cloud server; 2) There is only one round of communication between the client and the cloud server. When the dimensionality increases, the computational cost of our protocol approaches the computational cost of the original problem.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750Y (2022) https://doi.org/10.1117/12.2660267
Noise is an interference in the inversion process. To analyze its influence on the model selection in the inversion, this paper selects a uniform seabed model with a layered structure through simulation and uses the fast field method (FFM) to conduct acoustic field calculate. The transmission loss (TL) calculated by the acoustic field is added to the Gaussian noise as the research object, and the acoustic speed, density and acoustic speed attenuation are the inversion objects. The inversion results show that, after adding noise, the inversion method established in this paper can accurately achieve model selection, and the Root Mean Square Error (RMSE) within 0.95, it is verified that the inversion method still has strong anti-noise and accuracy under the influence of noise.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750Z (2022) https://doi.org/10.1117/12.2660699
Taking a chemical relocation plot in Shandong Province as an example, the investigation of trichloromethane concentration in soil was carried out by the systematic zonal distribution method, and the interpolation simulation of trichloromethane concentration in soil was carried out by the radial base model, IDW model and Tyson polygon model interpolation method respectively through statistical analysis of the survey sampling and detection results, comparing the simulation prediction accuracy of each interpolation model and different parameters, and the spatial distribution of trichloromethane concentration obtained by different calculation methods. The spatial distribution characteristics of contaminated soil volume and trichloromethane concentration obtained by different calculation methods were compared, and on this basis, the optimal model was selected for the spatial distribution simulation study of trichloromethane. The results showed that: the concentration of trichloromethane in soil was correlated with the depth of soil layer, and the concentration decreased gradually with increasing depth, the highest value of detected concentration was 5.96 mg/kg, and the average concentrations of trichloromethane in soil layer from top to bottom at three depths were 5.96, 1.67 and 1.18 mg/kg, respectively; there was a certain spatial correlation of trichloromethane concentration in soil, and by comparing the IDW (inverse distance weight) model can better simulate the spatial distribution characteristics of trichloromethane.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247510 (2022) https://doi.org/10.1117/12.2659676
Equipment with simulation training capability makes the equipment can be used in a real environment to carry out simulation training, to achieve the "equipment only war" to "equipment war training one" use mode change; to a typical optoelectronic detection equipment as the object, built the universal design framework of the equipment simulation training system based on LVC simulation, the composition and function of signal simulator and simulation training software are studied, and the simulation training flow is designed.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247511 (2022) https://doi.org/10.1117/12.2659371
The segmentation of fundus vascular images is a key step in retinal image processing and analysis. It plays an important role in the early prevention and diagnosis of eye diseases and blood system diseases. In view of the low accuracy of retinal blood vessel segmentation caused by the complex features of fundus images, the presence of soft, hard penetrant and bleeding noises, the low contrast between retinal blood vessels and background, uneven changes in width and curvature, small shape and blurred edges, this paper proposes a new network model RCU-Net for retinal vessel segmentation. Specifically, this paper proposes CBAM-Conv convolution block and RA-Conv convolution block based on visual attention mechanism (CBAM) and residual connection idea, which can make the network better use of the information at the spatial and channel levels. Experiments were carried out in public datasets DRIVE and CHASE_DB1. The accuracy, sensitivity, F1-Score and AUC of RCU-Net network on DRIVE datasets reached 96.75%, 82.02%, 81.45% and 98.33% respectively. Compared with the classical U-Net network, the proposed model can effectively improve the performance of retinal vessel segmentation.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Hua Luo, Ke Zhang, Junyun Shang, Lili Zhang, Na Yang, Jing Guo
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247512 (2022) https://doi.org/10.1117/12.2659583
The traditional hand-eye calibration needs to take the calibration target image acquired by the vision system and the corresponding robot pose information as input, the process is complicated, and there is a risk of calibration failure when the data are not corresponding. To solve these problems, a fast hand-eye calibration method is proposed. According to the principle that a certain point in space has the same three-dimensional (3D) coordinates in its coordinate system after the motion of the coordinate system, to control the end of the robot to carry out multiple translation and rotational motion, the traditional hand-eye calibration model AX=XB is simplified into AX=B. The hand-eye calibration matrix can be quickly solved only by the transformation between the camera coordinate system and the calibration target coordinate system. The experimental results show that the proposed method is simple, fast and high accuracy. It can be applied to eye-in-hand system and eye-to-hand system.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247513 (2022) https://doi.org/10.1117/12.2659679
In this paper, we design a convolutional neural network based on the ideas of depthwise separable convolution and inverted residual module. The scaling factor of BN layer is used as a measure for channel pruning of the network model to compress it. By analyzing the layer-by-layer pruning process of conventional convolution, the layer-by-layer pruning method with depthwise separable convolution and inverted residual structure is proposed to prune the channels of the network model, and finally, the channel pruning strategy of classification simplification network is developed. Tests on the selected dataset showed that the classification accuracy of the pruned and fine-tuned network model is 97.7% when the pruning rate is 0.7.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247514 (2022) https://doi.org/10.1117/12.2659345
IFC (Industry Foundation Classes) is a standard format for information exchange developed by Building SMART, dedicated to the collaborative work of various software in architectural design, construction and operation and maintenance. With IFC standard for various BIM (Building Information Modeling), the software provides a unified data structure and file exchange format for data exchange. However, lacking formal rigidity, data exchange is often arbitrary and prone to errors, omissions, and misrepresentations. This study applies the machine learning technique LightGBM to examine BIM elements and IFC The accuracy of the mapping between classes is extracted through feature engineering, and the BIM model element detection model is constructed. By using the BIM model training set for training, the results show that our model is more than 97.8 % accurate. And compared to the popular machine learning models, our model has higher performance.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Hua Luo, Ke Zhang, Junyun Shang, Meng Cao, Ruifeng Li, Na Yang, Jun Cheng
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247515 (2022) https://doi.org/10.1117/12.2659724
In the process of intelligent manufacturing, all kinds of complex workpieces in aerospace, automobile and other fields need to be measured and identified with high precision, so that industrial robots can sort or assemble the workpieces. The structure of the workpieces is complex, the surface texture is weak, and they are scattered and stacked on the automatic production line, so there are some problems such as low accuracy of three-dimensional (3D) measurement and positioning and low efficiency. To solve these problems, a high precision positioning method based on robot-driven 3D measurement is proposed. Firstly, the 3D point cloud data of the complex workpieces is obtained from the structured light 3D measurement device, and then the point cloud data is processed by the sampling consistent initial registration algorithm (SAC-IA) and the iterative nearest point algorithm (ICP). Through the rough estimation and accurate solution of the position and attitude of the workpiece, the 3D attitude of the workpieces in the coordinate system of the structured-light 3D measurement device is obtained. Finally, the spatial pose solution algorithm is used to calculate the 3D attitude of the workpieces in the robot coordinate system and guide the robot to grasp automatically. The experiments show that the grasping position error is 0.34mm, and the grasping angle error is 0.36°. It can accurately measure and identify the point cloud target, calculate the 3D attitude of the complex workpieces, and accurately guide the robot to grab the workpiece automatically, which can be popularized and applied in the industry.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Zhongrong Xu, Lu Wang, Lei Zhang, Haiwen Chen, Lulu Cui, Shiying Wang
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247516 (2022) https://doi.org/10.1117/12.2659698
The efficient extraction model of typical Chinese urban buildings is one of the important research contents to provide technical support for urban planning. However, the different architectural styles, a large number of vegetation and shadows in typical Chinese cities make the extraction of buildings with high errors. To address the problems of slow convergence speed and rough edge segmentation of the K-net combined with DeepLabv3 model on the remote sensing images of typical urban buildings in China, the Swin_ASPP_Knet network is proposed, which adds a multi-scale feature fusion module to enrich the feature information, and uses Swin Transformer to replace Resnet as the backbone network to extract more accurate feature information and accelerate the model convergence speed. After the comparative analysis of the experiments, the results show that the Swin_ASPP_Knet network outperforms K-Net combined with DeepLabv3 and Unet networks in the extraction task of typical urban building datasets in China, with mIoU and PA reaching 86.38% and 95.27%, respectively. To verify the generalization ability of the model, experiments were also conducted in the WHU building dataset where mIoU and PA reached 93.65% and 98.00%, respectively. Compared with the original model, the Swin_ASPP_Knet in both public building datasets reduced the model training time by half, and achieved good results in edge extraction of Chinese urban buildings with complex backgrounds, which met the requirement of efficient and accurate extraction of typical urban buildings in China.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247517 (2022) https://doi.org/10.1117/12.2660329
The DCC (Digital Command Control) protocol is a standard used in model railroading to control individual locomotives or accessories by modulating the track supply voltage. The commands to the trains are generated by a DCC unit that we are going to implement in the FPGA of the Nexys card, in the form of a mixed hardware/software system Thanks to a user interface created using the buttons of the Nexys card, the FPGA must generate a digital control signal. This command must then be amplified in current using a Booster card, to obtain a signal powerful enough to be sent on the rails and then be decoded by the locomotives.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247518 (2022) https://doi.org/10.1117/12.2659644
In various text classification tasks, models based on bidirectional encoder representations from transformers (BERT) have been widely used, but they may lack the semantic knowledge of phrases and the grammar knowledge of the text. In this paper, we propose a joint model (ERNIE2.0-DICNN) based on enhanced representation through knowledge integration (ERNIE2.0) and deep isometric convolutional neural networks (DICNN). Experiments on THUCNews, Weibo, and IMDB datasets demonstrate the higher accuracy of the proposed method.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247519 (2022) https://doi.org/10.1117/12.2659601
Data distribution service (DDS) is a middleware API standard from Object Management Group (OMG), which transfers data using a publisher-subscriber model. The number of distributed nodes deployed in today's DDS communication system can reach tens of thousands, thus improving the efficiency of the communication system is important. In this work we present a structured text encoding strategy based on prior dictionary. This compression method has considerable compression effect on structured text such as HTML. In our evaluation, the average data compression rate is reduced by 7.07%, and the average system latency is reduced by 8.61% comparing to Zlib.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Jie Dong, Zijie Ma, Zhibin Zang, Xiangdong Chen, Mingduan Zhou
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751A (2022) https://doi.org/10.1117/12.2660316
Aiming at the problem of low efficiency and automation of CORS massive data precision processing, an automated batch processing method for CORS massive data is proposed. Based on the Bash programming language, an automated batch script program for CORS massive data is designed and built. Through the examples in the paper, it is verified that the script of the proposed method is feasible and effective and has certain practical value for the application of power CORS massive data precision processing technology.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751B (2022) https://doi.org/10.1117/12.2659714
With the rapid expansion of the scale of the university, the demand for faculty, classrooms and courses has greatly increased, and the university course timetable plan has increased geometrically. This leads to the problem of combinatorial explosion in the course schedule. In this paper, a divide and conquer based course timetabling algorithm is proposed. By reducing the space vector of course arrangement, the number of course arrangement combination schemes can be effectively reduced, and the optimal solution can be found in a huge space, and effectively suppress the combinatorial explosion. Practice shows that the algorithm can adapt to the needs of dynamic large-scale class scheduling.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Advanced Algorithms and Machine Learning Applications
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751C (2022) https://doi.org/10.1117/12.2660126
With the development and application of distributed cloud computing, the problem of assigning workflow tasks to computational resources has become more and more prominent. It involves multiple constraints and optimization objectives, and is a typical NP-hard problem. Existing evolutionary algorithms face local optimum and premature convergence problems. Considering these facts, we proposed a multi-objective evolutionary algorithm with elitism strategy (MOEAES) in this paper. To avoid local optimum, MOEAES uses a new crossover operator called Random Sub-Sequence Exchange Crossover (RSSEX), and it introduces a multi-population-based elitism strategy to accelerate the algorithm. Finally, experimental validation is carried out, which shows that MOEAES achieves performance improvement in terms of solution quality and convergence speed comparing to other methods.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751D (2022) https://doi.org/10.1117/12.2659363
Monte Carlo (MC) simulation is a powerful technic to valuate option, but few literature discuss option pricing under stochastic volatility model (SVM). In this paper, we apply MC method to price European options under SVMs. Firstly, given correlative coefficient, a formula generating norm distribution random variables is established. Then, MC scheme is proposed for pricing European options under Heston, Hull-White and Hyper-geometric models. Numerical experiments illustrate the efficiency and accuracy of MC algorithm. With large number of simulated paths and time partition, MC solutions become stable. The proposed MC method can be extended to general option pricing, such as local stochastic volatility models and so on.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751E (2022) https://doi.org/10.1117/12.2659307
The bus was the priority travel mode for most citizens in the urban public transportation network. The research of bus transfer and query algorithm, the design of a system that can run on the Android platform and can query bus information promptly and accurately has become a realistic application requirement. In this paper, in terms of the issue of public transport transfer query and transfer, a weighted directed graph was adopted to describe the bus network. The public transport transfer matrix was constructed by a weighted directed graph based on the bus network. According to the actual test, the bus route display, the query of the route and the query of the transfer scheme could be achieved by the designed algorithm, which proved that the algorithm was effective.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751F (2022) https://doi.org/10.1117/12.2659668
Aiming at the optimization of emergency evacuation paths for personnel in terminal emergencies, a path optimization method based on minimizing the cumulative danger is proposed. Taking the minimum cumulative danger degree and the minimum total evacuation time as the objective functions, a multi-objective path optimization model for emergency evacuation is established; an improved Dijkstra-GA path optimization algorithm is proposed to solve the problem, and an emergency evacuation path optimization scheme is obtained; an airport terminal is taken as an actual case, using Anylogic software to simulate the obtained scheme, and compared with the results of the traditional Dijkstra algorithm. The simulation results show that the evacuation time of the evacuation scheme solved by the improved Dijkstra-GA algorithm is 378s, which is 229s higher than that of the traditional Dijkstra algorithm. The feasibility of the path optimization scheme based on the improved Dijkstra-GA algorithm is verified.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751G (2022) https://doi.org/10.1117/12.2659348
Aiming at the problems encountered by the traditional ant colony optimization in the time optimal path planning, the method of transforming the ant-cycle system model is studied. In the optimization process, it is coupled with the algorithm model by quantifying the node turning time, the pheromone update rule is adjusted accordingly, and the computational efficiency is taken into account. The improved mathematical model is simulated by MATLAB software, and the effectiveness of the improved ACO model is verified by comparing the optimization results of the standard map traveling salesman problem with the classical ant colony. The main variable parameters are selected, and the optimal combination is preliminarily screened out through the simulation analysis of their sensitivity to the results.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751H (2022) https://doi.org/10.1117/12.2659634
Aiming at the shortcomings of the seagull optimization algorithm in the iterative process, such as slow convergence speed and easy to fall into local optimum, an improved seagull optimization (CNSOA) algorithm based on nonlinear convergence factor and mutation using the Cauchy operator is proposed. The tent chaotic mapping strategy is used to initialize the population that make the seagull population more uniformly distributed in the search space. In the process of seagull migration, a nonlinear convergence factor is used to guide the seagull to seek optimization, so that the algorithm has better search ability. The Cauchy mutation perturbation strategy is adopted to make the algorithm better jump out of the local optimum. Finally, 9 benchmark test functions are used to test the CNSOA, and the results are compared with the SOA and 5 famous algorithms. The experimental results show that the CNSOA performs better in convergence speed and jumping out of the local optimum.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751I (2022) https://doi.org/10.1117/12.2659479
As a widely applied data hiding method, least significant bit (LSB) algorithm is a spatial steganography which uses the lowest bits of each pixel’s grayscale value to hide messages. In spite of its high capacitance, high load rate and imperceptibility is remarkable, the lack of robustness is the fatal problem which prevents the LSB algorithm’s further application. This paper proposes a new high-robustness LSB method based on the Reed-Solomon (RS) code and provides higher immunity of attack with slight cost of imperceptibility and capacity. This algorithm exploits the considerable error correction ability of RS-code to encode the watermark message before embedding them to the carrier image. Despite the size of the encoded message increases times to the original, this method makes self-correction of encrypted images feasible, which would provide more superiority among the algorithms that require carrier image as reference to extract watermark message. This paper provides the testing results of the LSB algorithm based on RS-code in Matlab and verifies the immunity of both the proposed method and original method against noise and rotation attack, and finally determines that this algorithm is able to substantially enhance the robustness of encrypted image under a certain range of noise density.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751J (2022) https://doi.org/10.1117/12.2659383
With the development of the Internet of Vehicles (IoV), Parked Vehicle Edge Computing (PVEC) has gradually attracted people's attention. Parked vehicles (PVs) can be exploited as a supplementary computing resource to Mobile Edge Computing (MEC). Using the onboard resources of parked vehicles can effectively improve the security of the blockchain-based internet of vehicles system. However, there are multiple MEC nodes belonging to different service providers and multiple parked vehicles in an area. Due to individual rationality, parked vehicles and MEC nodes will not provide services for free. Therefore, in this paper, we study the interaction of MEC nodes and parked vehicles in blockchain-based parking vehicle edge computing and model it as a two-stage Stackelberg game to optimize the utility of MEC nodes and Parked vehicles. Specifically, we treat the parked vehicle as a leader, set the price of computing resources, treat the MEC node as a follower, and determine the demand for computing resources through the price of the leader. We use ADMM to optimize their utility function to obtain a computing offloading scheme that maximizes system utility. Simulation results show that our scheme can maximize the utility of MEC nodes and Parked vehicles and maximize social welfare.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Yaping Yin, Haitao Li, Shilin Sun, Zhen Liu, Yichuan Wang
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751K (2022) https://doi.org/10.1117/12.2659342
Sonar pulse detection and recognition is an important research direction of national marine construction. Traditional target detection and recognition methods have insufficient feature extraction capabilities and high time complexity. To solve this problem, this paper makes full use of the strong feature expression capabilities of deep neural networks. Based on the mainstream target detection networks including Faster RCNN, SSD, and YOLOv3, the sonar pulse detection and recognition method based on deep learning are deeply studied and verified on the pulse signal generated by simulation. The experimental results and analysis show that the average detection accuracy of the YOLOv3 network for sonar pulse signals can reach 92.35%, and the detection time of a single pulse signal power spectrum is only 0.018 seconds. Compared with Faster RCNN and SSD, YOLOv3 has better practicability and robustness in the field of sonar pulse detection and recognition.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751L (2022) https://doi.org/10.1117/12.2659684
Density Peak Clustering (DPC) has become the hottest unsupervised clustering algorithm today since it does not need any iteration and can find cluster centers at one time. In order to further improve the robustness and usability of DPC while maintaining its accuracy, this paper proposes a novel model named Density Peak Clustering Based on Sparrow Search Algorithm and Improved Shared Nearest Neighbor, SSA-SNN-DCP in short. It can deeply mine the hidden spatial angle features of data points and use the sparrow search algorithm to find the right clustering threshold to avoid the influence of artificial prior knowledge. First, an improved SNN similarity measure based on geometric mean and cosine similarity is proposed, amplifying the shared neighborhood compensation mechanism. Then, a cluster center selection strategy based on SSA is designed to realize automatic cluster center election. Finally, the performance of the algorithm is verified using 3 synthetic datasets and 3 UCI real datasets. The experimental results show that SSA-SNN-DPC has a better performance whether it is on spherical clusters, multi-density overlapping clusters or helical clusters.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751M (2022) https://doi.org/10.1117/12.2659693
Aiming at the problems of low solution accuracy, slow convergence speed and easy to fall into local optimization, a multi-strategy fusion improved puzzle optimization algorithm (IPOA) is proposed. Improved Circle mapping strategy, dwarf mongoose optimization (DMO) strategy, and adaptive t-distribution variation strategy were used to increase the diversity of puzzle populations, improve global search capabilities and local development performance. Through the analysis of multiple benchmark function simulation experiments and PID controller parameter optimization problems, the results show that the IPOA algorithm has high optimization accuracy, fast convergence speed and effectiveness in solving practical engineering problems.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751N (2022) https://doi.org/10.1117/12.2659988
In order to solve the problem of steel surface defect detection, an improved algorithm based on YOLOv5 is proposed. EIOU loss is used to replace the original GIOU loss function, and the attention mechanism SE module is added to the network model to strengthen important characteristic channels. By setting different training parameters in the steel defect set for multiple rounds of testing, the results show that under different parameters, the improved YOLOv5s model can detect steel surface defects with the mAP value of 86.9%, which is 8.7% higher than the original model. Compared with traditional steel surface defect detection methods, the proposed algorithm can detect the types and locations of steel surface defects more accurately.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751O (2022) https://doi.org/10.1117/12.2659385
In the process of traditional weapon-target assignment, there are two critical issues that have not been well-studied: (1) the waste of firepower resources and (2) the lack of description of the relationship between targets. Towards this end, we propose an improved weapon-target assignment model, which combines the damage probability, the weapon resource consumption and the relationship coefficient between the targets. On this basic, we design a multi-objective whale optimization algorithm based on grid division (GDMOWOA). Specifically, the algorithm uses the grid partitioning method to sort the population non-dominated, selects the optimal individual by calculating the grid number and density, and introduces an external pareto archive to maintain the population diversity. Simulation experiments are conducted to verify the rationality and effectiveness of our solution. The results show that our algorithm has better effectiveness and superiority compared with other classical algorithms, and can effectively solve the problem of weapon-target assignment.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Jianning Li, Junzhao Zhang, Xinyu Zhang, Shuai Wang
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751P (2022) https://doi.org/10.1117/12.2659641
In the construction process, wearing safety helmet can effectively reduce the injury caused by falling objects and hard objects. In order to detect the wearing of safety helmet during the construction process, a lightweight safety helmet wearing detection algorithm YOLOv5-CS is proposed and applied to the embedded equipment at the construction site. The algorithm is based on the improvement of YOLOv5 target detection algorithm. The improved CAShuffle module is used to form the backbone feature extraction network, which reduces the size of the original algorithm and ensures the detection accuracy of the algorithm; DIoU as the loss function, which makes the algorithm more conducive to detecting small targets such as hard hats; modify dataset, add the dataset on the basis of the original dataset and mark the safety helmet and the anchor frame of the person separately, and judge whether the worker wears the safety helmet by analyzing positional relationship of the identified anchor frame, use the algorithm combining K-means and genetic algorithm to predict the anchor frame of the dataset, and use Mosaic algorithm to enhance the data. The results show that the parameter quantity of YOLOv5-CS algorithm is 2.4M, less than one third of YOLOv5; The average precision decreased by only 4.2%; 640x640 pictures can be recognized at 31 frames per second on Jetson Nano, a embedded platform.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751Q (2022) https://doi.org/10.1117/12.2659858
Aiming at the vehicle routing problem with time window constraints, this paper discusses the vehicle routing optimization problem among multiple warehouses, multiple products and multiple customers. In this paper, a multi-objective model of vehicle path optimization based on the improved simulated annealing algorithm is established by adding memory function and setting monotonous heating up and other real-time optimization strategies. The optimization objectives include total transportation cost, total driving distance and total driving time. In order to prove the effectiveness of the proposed model, the optimization results of standard genetic algorithm, random search and the algorithm in this paper are compared on the randomly generated dataset. The simulation results show that the algorithm designed in this paper is fast and efficient in all four cases.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751R (2022) https://doi.org/10.1117/12.2659334
Container loading is an important step in the process of cargo transportation and belongs to the NP-hard problem. All different sizes of containers are loaded into different sizes of containers and the space utilization of containers is maximized. In this paper, we mainly study the loading layout of a single container, and establish a mathematical model with the volume, load capacity, cargo direction and stability of the container as constraints and the utilization rate of container space as the final goal. Space partitioning, simple block generation heuristic algorithm and depth-first search algorithm are used to solve the model. The feasibility and applicability of the algorithm were verified by specific cargo data. The average volume utilization rate of three foreign trade containers is higher than 89%, which achieves the ideal crating effect and proves the effectiveness and feasibility of the algorithm.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751S (2022) https://doi.org/10.1117/12.2659407
With the in-depth development of intelligent transportation, traffic sign recognition has attracted widespread attention as an essential part of intelligent transportation. This paper studies several machine learning methods for traffic sign recognition. Through comparative analysis, it is found that Convolutional Neural Network (CNN) is superior to Support Vector Machine (SVM) and K Nearest Neighbor (KNN) methods in recognizing traffic signs. And adding Gaussian noise to the image data for enhancement can further improve the accuracy of applying a Convolutional Neural Network to identify traffic signs. The accuracy of applying a Convolutional Neural Network to identify traffic signs is 99.2%. After adding Gaussian noise with a mean of 0 and a standard deviation of 1 to the image set, the accuracy of applying a Convolutional Neural Network to identify traffic signs was increased to 99.6%. We also compared the CNN-based traffic signs recognition experiment in this paper with the experiments of two other scholars. Our experiment has higher accuracy in a particular data range and environment.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Qizhi Zhang, Qiming Yu, Yang Xiao, Ziyu Feng, MengWei Wu
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751T (2022) https://doi.org/10.1117/12.2659372
Domain Generation Algorithm (DGA) is an algorithm used by malware to generate a large number of domain names, or DGA domains, on a regular basis. This ancient but ever-active technique is the key weapon on which central structure botnets rely. In recent years, many botnets have adopted the DGA algorithm's domain transformation technique to evade detection and blocking, making it extremely difficult for security personnel to detect DGA domain names. In this article, we use the domain names of Alexa's top one million global websites as a white sample. For the DGA sample data, we take the open data of 360NetLab as the black sample to form the data set of this paper. We use 2-Gram model for feature extraction, and construct the DGA domain detection model based on the LightGBM algorithm. Experimental results show that the accuracy of our model is higher than 98%, and compared with the current common classification models, our model has better performance in both time and space.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751U (2022) https://doi.org/10.1117/12.2659603
In order to improve the efficiency of model learning and the accuracy and precision of detection results in network intrusion detection, an algorithm based on improved immune cloning algorithm to optimize neural network structure and parameters is proposed. The raw data is dimensionally reduced using principal component analysis and used as model input data. The neural network structure and parameters are used as immune antibodies. When the antibody group evolves slowly in the process of immune cloning, the antibodies with high affinity are used as booster vaccines to inoculate to achieve the optimal selection of neural network structure and parameters. The experimental results show that compared with the unoptimized detection model, the detection model optimized by traditional immune clones and the detection model improved by other optimization algorithms, the method improves the accuracy, precision and false alarm rate.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751V (2022) https://doi.org/10.1117/12.2659682
In order to further improve the exploration and exploitability of Differential Evolution (DE) algorithm, Bessel mutation strategy based on grid entropy is proposed to improve iLSHADE-RSP algorithm. The new algorithm is called BiLSHADE-RSP. It has the following features: (1) A grid entropy for observing the convergence of the population is proposed to be used as a signal of applying Bessel mutation strategy; (2) A mutation strategy based on Bessel curve is designed to enhance the exploitability of DE. BiLSHADE-RSP is compared with five algorithms, namely iLSHADE-RSP, LSHADE-RSP, L-SHADE, DCDE and FNADE. We use CEC2017 test suit to verify the performance of the proposed algorithm. Experimental results show that the improvement in BiLSHADE-RSP algorithm has achieved the best effect on complex test functions, and its solution accuracy is better than the current popular DE variants.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751W (2022) https://doi.org/10.1117/12.2659700
Aiming at the problem that there are many deficiencies in the existing dressing recommendation service, a big data dressing recommendation service system based on the optimization algorithm of BP neural network combined with the ant colony is designed. First, based on the weather data of the city where the recommended service object is located, the weather data set is constructed, and then, the dress recommendation model is constructed, based on the current weather data set, the dressing index is calculated, and finally, the push set is generated according to the dressing index recommendation table to recommend the dressing to the user. Based on the real data collected by the smart light pole, the local weather data set is constructed to achieve the recommendation results, and the recommendation results show that the recommendation algorithm improves the recommendation accuracy. Addresses issues such as inaccurate recommendation results.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751X (2022) https://doi.org/10.1117/12.2660127
With the continuous development of e-commerce websites, online shopping has become an indispensable part of life. However, the huge amount of product information makes customers lost and weakens the ability to obtain the expected information. In recent years, sorting algorithm has been applied in recommendation system to recommend information and products to customers according to product information and customer behavior patterns. Based on a large number of historical sales commodity data and customer data, combined with LightGBMRanker algorithm and feature engineering processing, this paper establishes a commodity recommendation and sorting model. And judge the accuracy of the sorting model according to the subsequent purchase data of customers. The results show that our model can effectively provide customers with desired product recommendations, and the prediction accuracy is higher than other algorithms. Specifically, the maximum map@12 value of our model is 0.0284, which is 26.22% and 22.41% higher than SVM algorithm and LSTM algorithm respectively. In addition, the ranking of important features that affect product sorting and recommendation is given in this paper, and some constructive guidance is obtained.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751Y (2022) https://doi.org/10.1117/12.2659964
This paper studies the evaluation model of higher vocational English teaching effect based on particle swarm optimization and support vector machine classification. Based on the relevant theories, the model constructs a multi index evaluation system of English teaching effect in higher vocational colleges, which takes teachers and students as the main body. This model improves the accuracy and efficiency of English teaching evaluation in higher vocational colleges and meets the requirements of English teaching evaluation in higher vocational colleges. The experimental results show that this method has a good effect on the evaluation of English teaching in higher vocational colleges. The average evaluation accuracy is 97.1%, the evaluation time is short, and the test time is as low as 6ms.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124751Z (2022) https://doi.org/10.1117/12.2660279
As a basic source of energy in China, the rise in the price of coal has a bearing on the development of the Chinese economy. In view of the influencing factors of thermal coal price, the prophet algorithm and long term memory algorithm are proposed to predict the price of thermal coal. Based on the characteristics and principles of these two algorithms, and according to the characteristics of thermal coal, the parameters suitable for the experimental object are selected. Through the prediction of the historical data by using these two algorithms, analyzing and comparing the error of the prediction results with the actual value. Following this, a time series algorithm more suitable for thermal coal is obtained.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247520 (2022) https://doi.org/10.1117/12.2659592
Machine learning can be applied to cancer diagnosis, neurosurgery, radiotherapy, and so on. Its high accuracy and efficiency can improve the hospitals’ overall efficiency and save diagnostic time. This paper aims to provide an algorithm for extracting critical information from x-ray medical images. The extraction process was divided into two parts: 1) the pixel coordinates of cervical vertebra nodes via automatic localization and 2) patient-related data using OCR (optical character recognition). The extracted pixel coordinates and the text information in the image were then validated. Compared with manual processing of medical images, the proposed algorithm in this study provides higher efficiency, which can better serve doctors to evaluate patients’ medical images for further diagnoses. In the future, more and more medical images can be analyzed in a more intelligent and efficient way.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247521 (2022) https://doi.org/10.1117/12.2659373
With the rapid development of the internet, the spread of fake news in the media is vast and fast, becomes a social phenomenon. The proliferation of fake news has a devastating impact on the country and society. Many scholars are looking for automated methods to detect fake news. For saving computing resources, fake news detection is often solved by using machine learning algorithms. However, there are many hyperparameters in machine learning algorithms. And different hyperparameters significantly impact the model, which leads to significant differences in the results of the same model for specific problems. Finding a way to obtain stable and accurate hyperparameters has become a research direction of many scholars. This paper first analyzes the characteristics of fake news and then uses genetic algorithm to assist the decision tree model in finding hyperparameters. Compared with random search and traversal hyperparameter search within a specific range, this combined method is faster and more accurate. The results verify the high accuracy of the proposed method. The proposed model can improve the classification performance and is a compelling new method to solve fake news detection.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247522 (2022) https://doi.org/10.1117/12.2660269
The online allocation of course resources is a part of the field of cloud computing. Studying the allocation of digital story course resources through cloud computing architecture can not only meet the requirements of cloud computing resource allocation, but also improve the utilization efficiency of digital story course resources. This paper mainly proposes an application-oriented resource allocation algorithm based on deep learning. This online allocation algorithm mainly quantifies the characteristics of the data, and more accurately creates the server resources required by the digital story course. In addition, the workload prediction model is integrated into the online allocation algorithm, which can Ensure that curriculum resources and teaching are more matched, so as to give full play to the value of resource use.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Proceedings Volume Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 1247523 (2022) https://doi.org/10.1117/12.2661078
In the process of facial expression recognition by convolutional neural network, aiming at the problem that the complex background interferes with the extraction of expression features, a simple face cropping strategy is proposed. First, the critical facial expression regions are calculated by face alignment and landmarks detection, thus the background influence outside the facial expression region is reduced, and then the convolutional neural network is used to further extract expression features and enable expression classification. The experimental results show that the facial expression recognition effect is significantly improved by the proposed method, and the recognition accuracy on the facial expression datasets JAFFE and CK+ reaches 90.48% and 96.67%, respectively.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.