Traffic events can be carried out by traffic surveillance cameras and are paramount in ITS. Adverse weather conditions restrict the camera’s function by reducing the qualities of videos and images and increase the probability of the false detection to target vehicles. So, the research for vehicle detection under typical complex weather conditions is crucial for the development of ITS.
A vehicle detection methodology composed with image enhancement+Yolov5 network under typical complex weather conditions is developed in this paper. First, the specified three image enhancement methods are researched, and their efficacy is evaluated by objective evaluation methods to adverse weather conditions. After that, appropriate enhancement algorithms are chosen for different degraded images, the enhanced images are put into Yolov5 network for training and the high detection precision has achieved on the validation dataset. Then, the comparison research is performance among Faster R-CNN, YoLov3 and the methodology proposed in this paper. It is found that the methodology has higher detection precision and lower time cost, and offers a better balance between the accuracy of detection and the velocity of execution in the three approaches.
Optical projection tomography (OPT) is a three-dimensional (3D) imaging technique for biological samples, capable of visualizing tissues, embryos, and organs within 1mm to 10 mm scale. Filtered back projection (FBP) is an extensively used 3D reconstruction algorithm for OPT with dense sampling data from all view angles. In-vivo OPT can reduce the inspection time by using equally-spaced sparse angle projections to mitigate the side effects of phototoxicity and anesthetics. This work compares the reconstruction results of the sparse-angle OPT using different algorithms, including the FBP algorithm and two kinds of compressive sensing (CS) algorithms with different projection numbers. We also build up a testbed of OPT to verify these algorithms using experimental data. It shows that the CS algorithms result in reconstructed images with fewer artifacts compared to the FBP algorithm. Especially, the advantage of CS algorithms over FBP algorithm becomes more obvious as the projection number is reduced.
KEYWORDS: Color, Neural networks, Deep learning, Education and training, RGB color model, Detection and tracking algorithms, Data modeling, Object detection, Histograms, Feature extraction
Color features play a unique role in vehicle recognition. The color recognition algorithms based on deep learning neural networks are studied in this paper. A color recognition experiment is carried on to some typical deep learning neural networks, the result of the experiment proves that Yolov5 has faster training speed and higher accuracy for vehicle color recognition, so Yolov5 is chosen for the color recognition. The structure of yolov5 is optimized by adding C2f module replacing C3 module and adjusting the parameters of HSV color space when it is applied to identify 8 typical vehicle colors using BIT Vehicles data set. The modified Yolov5 make the accuracy of vehicle color recognition improved effectively to the complex color vehicles and part covered vehicles comparing with original yolov5 network.
Color is one of the most features of vehicles and can be used for vehicles recognition. Deep learning has greater advantages for vehicle color recognition over traditional algorithms. In this paper, we present a vehicle color recognition method using GoogLeNet with Inception v1. Inception v1 increases the width and depth of the network, reduces the parameters to save computing resources and uses sparse matrix to avoid redundancy of traditional neural network as well. We use a publicly dataset to train and validate GoogLeNet and a self-made dataset to test the method. The method can recognize regular eight kinds of vehicle colors and the probability is stable at 90%-95%. Afterward, we have a discussion on how the impact of different datasets on the method as well as the possible reasons. In the future, we will combine GoogLeNet and Yolo network structure to research vehicle color recognition.
Optical projection tomography (OPT) is a tool used for three-dimensional imaging of millimeter-scale biological samples. For higher image quality, new methods will need to be researched for OPT imaging systems. To make full use of the advantages of light polarization, an OPT image system with a polarization device was built, which can provide polarized projection data. The optimum polarization angle of the polarization device was acquired by experiments. At the optimum polarization angle, the high quality polarized projection data of samples were obtained, the reconstructed tomographic images got more details of samples and the influence of the stray light was eliminated. FSRCNN(Fast Super-Resolution Convolutional Neural Network)based on deep learning was applied for the SR reconstruction of tomographic images. SR tomographic images were assessed by the metrics of image quality and subjective observation. The outline and details of the samples were considerably represented in three-dimensional images reconstructed by SR tomographic images. So, polarization technology and FSRCNN can complement the performance of OPT imaging systems, and enhance imaging ability in the micron range.
Since it was first presented in 2002, the Optical Projection Tomography(OPT) imaging system has emerged as a powerful tool for the study of a biomedical specimen on the mm to cm scale. In this paper, we present a rough and precise algorithm to further improve OPT image acquisition and tomographic reconstruction. The rough and precise algorithm combines the merits of the binarization process and the maximum correlation coefficient, and can accurately correct the displacement of the rotation axis. The tomographic images corrected by the rough and precise algorithm have higher image quality in the simulation experiments and specimen experiments. The reconstructed 3D images based on tomographic images can restore the original specimens. Thereby, the rough and precise algorithm contributes to increasing acquisition speed and quality of OPT data. More work should be performed to better understand and amend the rough and precise algorithm by abundant specimen experiments.
Background is assumed to be uniform usually for evaluating the performance of thermal imaging systems, however the impact of background cannot be ignored for target acquisition in reality, background character is important research content for thermal imaging technology. A background noise parameter 𝜎 was proposed in MRTD model and used to describe background character. Background experiments were designed, and some typical backgrounds (namely lawn background, concrete pavement background, trees background and snow background) character were analyzed by 𝜎. MRTD including 𝜎 was introduced into MRTD-Channel Width (CW) model, the impact of above typical backgrounds for target information quantity were analyzed by MRTD-CW model with background character. Target information quantity for different backgrounds was calculated by MRTD-CW, and compared with that of TTP model. A target acquisition performance model based on MRTD-CW with background character will be research in the future.
The impact of nature environment on the synthesized performance of thermal imaging systems was researched comparing with the targeting task performance (TTP) model. A nature background noise factor was presented and introduced into the minimum resolvable temperature difference channel width (MRTD-CW) model. The method for determining the nature background noise factor was given. A information quantity model based on MRTD-CW model was proposed to evaluate the impact of nature environment on the synthesized performance of thermal imaging systems. A normalized parameter was introduced into the information quantity model. Different background experiments were performed, and the results were analyzed and compared with those of TTP model.
Based on the aliasing theory of focal plane array (FPA) thermal imaging systems, aliasing as noise (AAN) method and under-sampling system evaluation model based on information theory (EMIT) were analyzed. The aliasing was treated as one kind of noise and introduced into minimum resolvable temperature difference (MRTD) model, and the integral expressing of aliasing signal (IAS) method was proposed to evaluate the impact of aliasing on the performance of thermal imaging systems. IAS method was proved to be able to describe the impact of aliasing grade on the system performance effectively. The three MRTD models with different aliasing evaluation methods were researched contrastively by MRTD test experiment and target detection probability experiment. The MRTD model with IAS method was proved to evaluate the performance of under-sampling thermal imaging systems effectively, and could be used in the performance prediction of thermal imaging systems. In the future, the precision of IAS method will be researched further.
Discrete sampling is one of the important characteristics of thermal imaging systems and has an important effect on the
target acquisition performance. Based on the study of discrete sampling theory, several discrete sampling evaluation
methods are analyzed, and the aliasing caused by under-sampling is considered as a noise of thermal imaging systems.
The aliasing noise is introduced into MRTD model with system noise. MRTD model with aliasing noise is deduced, and
its validity is verified by simulation experiments. MRTD model with aliasing noise is introduced into MRTD channel
width model. It is future work that the impact of discrete sampling on the general performance of thermal imaging
systems will be researched by MRTD channel width model.
Background has a very important influence on point target operating range of infrared search and track system(IRST).
Based on the noise equation temperature difference(NETD) operating range model , a general noise parameter δ′ which
describes the influence of background and system noise on the operating range is presented, and a new operating range
model is founded including δ′. δ′ has clear meaning and gives explicit explanation for conventional detector-limited and
background-limited operating range model. The algorithm of δ′ based on the ratio of background region to target area is
proposed under natural sky background. The simulation shows that δ′ is valid for predicting point target operating range
of IRST.
Electronic devices are widely used in various industries, their temperature distribution cannot be obtained by traditional
test methods. In recent years, simulation softwares are used to simulate the thermal characteristics of electronic devices
and play a positive role on the reliability improvement, on the contrast, their validity cannot be verified. In this paper, the
chip temperature rise process is simulated by ICEPEAK software. Some factors that change thermal characteristics are
analyzed. The actual working temperature obtained by the thermal microscope is compared with the simulation
temperature. The validity of simulation temperature is tested and the relation is built between the actual temperature and
simulation temperature. Finally, it is pointed that thermal microscopes are the development direction on the electronic
devices design and reliability testing.
Traditional ACQUIRE models perform the discrimination tasks of detection (target orientation, recognition and
identification) for military target based upon minimum resolvable temperature difference (MRTD) and Johnson criteria
for thermal imaging systems (TIS). Johnson criteria is generally pessimistic for performance predict of sampled imager
with the development of focal plane array (FPA) detectors and digital image process technology. Triangle orientation
discrimination threshold (TOD) model, minimum temperature difference perceived (MTDP)/ thermal range model (TRM3)
Model and target task performance (TTP) metric have been developed to predict the performance of sampled imager,
especially TTP metric can provides better accuracy than the Johnson criteria. In this paper, the performance models
above are described; channel width metrics have been presented to describe the synthesis performance including
modulate translate function (MTF) channel width for high signal noise to ration (SNR) optoelectronic imaging systems
and MRTD channel width for low SNR TIS; the under resolvable questions for performance assessment of TIS are
indicated; last, the development direction of performance models for TIS are discussed.
KEYWORDS: Imaging systems, Modulation transfer functions, Optoelectronics, Charge-coupled devices, Eye, Human vision and color perception, Signal to noise ratio, Sensors, Video, Visual process modeling
The Square Integral Method based on the Minimum Resolvable Contrast (MRC) is to be introduced in this paper as
an evolution for the design and evaluation of optoelectronic imaging systems. It is well known that there exists an optimal
angle magnification which can make optoelectronic imaging systems and human eye matching optimally, so that
optoelectronic imaging systems can performance best. Based on MRC (Minimum Resolvable Contrast) and channel width,
a new method called Square Integral (SQI) method was presented for evaluating the general performance of a CCD imaging
system, and attaining the optimal angle magnification or optimal viewing distance. Results calculated with this method are
in good agreement with the experimental measurements. From the agreement between the practical use and the theoretical
predictions for the variation of CCD size, optical focus, luminance and human vision, it demonstrates that the SQI
method is an excellent universal measure for the optimal angle magnification and the performance of CCD imaging
systems.
KEYWORDS: Imaging systems, Minimum resolvable temperature difference, Modulation transfer functions, Thermography, Optoelectronics, Eye, Sensors, Human vision and color perception, Multichannel imaging systems, Signal to noise ratio
The Square Integral (SQI) Method based on MRTD (Minimum Resolvable Temperature Difference) is introduced for the
design and evaluation of thermal imaging systems. It is well known that there exists an optimal angle magnification which
can make optoelectronic imaging systems and human eye matching optimally, and make optoelectronic imaging systems
attaining the optimal performance. Based on MRTD and channel width, a new way called SQI method is presented for
evaluating the universal performance on thermal imaging systems, and attaining the optimal angle magnification or optimal
viewing distance. The method can give a rational description for the matching between thermal imaging systems and human
eye. Results calculated with this method are in agreement with experiment measurements quite perfectly. From the
coherence between measurement data and theoretical predictions with variable detector size, optics focus, luminance and
human vision, it appears that the SQI method is an excellent synthesized measure for the optimal angle magnification
and the performance of thermal imaging systems.
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.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
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.