This PDF file contains the front matter associated with SPIE Proceedings, Volume 12319, including the Title Page, Copyright information, and Conference Committee lists.

Recently, the deep learning technology has been successfully applied to many applications of optical metrology, e.g., fringe-pattern analysis, fringe denoising, digital holography, and three-dimensional shape measurements. However, deep neural networks (DNNs) cannot always produce a provably correct solution, and the prediction error cannot be easily detected and evaluated unless the ground truth is available. This issue is critical for optical metrology, as the reliability and repeatability of the measurement are of major importance for high-stakes scenarios. As most deep neural networks are driven by data completely and work without considering any physical principles, how to believe the prediction of the DNN in optical metrology is a big challenge. Inspired by recent successful Bayesian deep learning approaches, we demonstrate that a Bayesian convolutional neural network (BNN) can be trained to not only retrieve the phase from a single fringe pattern but also produce uncertainty maps depicting the pixel-wise confidence measure of the estimated phase. Experimental results show that the proposed BNN can quantify the reliability of phase predictions under conditions of various training dataset sizes and never-before-experienced inputs. We believe that a DNN that can provide confidence measure of the estimated phase is crucial to fringe-pattern analysis and it has great potentials for inspiring novel and reliable learning-based optical metrology approaches.

This paper proposed a Gamma effect correction method based on the probability distribution function (PDF) for suppressing phase nonlinearity error in fringe projection profilometry. In this work, the periodicity of phase with Gamma effect is first analyzed, and nonlinear wrapped phase is modeled as superposition of normal wrapped phase with a sinusoidal function. Afterwards, a series of reference phases with Gamma factors ranging from 1 to 3 is constructed and the corresponding PDF is calculated. Then the optimal precoding factor γp of projected fringe is obtained by applying Jensen-Shannon (JS) divergence matching between the PDF of measured phase and constructed reference phases. In simulations, the availability of proposed method is investigated, where RMSE decreases from 0.218rad to 0.016rad. In experiments, the turbine blade is tested and compared with calibrated values where RMS of measured deviations after correction has decreased from 0.269mm to 0.095mm. All the investigations have proved the high reliability of proposed method.

The article analyzes the composition of the Rotating Laser vertical error based on the vertical plane working principle of the rotating laser and build a three-dimensional coordinate system based on a vertical plane established by optical axes of two collimators which is symmetrically distributed in ±β° relative to horizontal axis. Using the sine and cosine theorems of spherical triangles, the formula of the inclination error of the vertical level plane of the rotating laser is deduced. Using the formation characteristics of cone-angle error, the formula of cone-angle error of horizontal level plane is used as of vertical rotating plane. Finally, we build a mathematical model of the vertical error

White light interferometry (WLI) provides noncontact, high-precision surface profiling and inspection for ultra-precision machining. This paper presented a signal time-domain mode-decomposition denoising based surface recovery algorithm for WLI. In this work, the captured correlogram is firstly decomposed into a series of modes with different central spectrums by means of the variation mode-decomposition (VMD), and the spectral component of each intrinsic mode can be derived through the Fourier transform. Afterwards, the noise existed in each spectral component is eliminated through windowed Fourier filtering (WFF), where the filtering threshold is decided by the ratio of spectral energy of intrinsic mode comparing with that of the correlogram. The denoised correlogram could then be extracted as the sum of filtered intrinsic modes. And the surface height isfinally retrieved through envelope peak location by applying Hilbert transform. The effectiveness of the proposed method on noise suppression is investigated under different levels of additive noises occurred on simulated correlograms. Moreover, a height step standard with calibrated values 1.762±0.010μm is further testified, where the measurement accuracy of the proposed method is totally verified.

Three-dimensional (3D) reconstruction plays an important role in intelligent manufacturing, industrial inspection and reverse modeling. The model accuracy of 3D reconstruction has important influence on the final product quality and reliability, and point cloud registration is the key to 3D reconstruction, whose registration accuracy directly affects the final 3D reconstruction accuracy. There are many researches on point cloud registration algorithms, but the existing point cloud methods often have the disadvantages of low accuracy and slow speed when registering large point clouds. To meet this challenge, it is proposed that a high-accuracy point cloud registration method by digital volume correlation (DVC). Firstly, source point cloud and target point cloud are down sampled by voxel grid filter. Subsequently, the intrinsic shape signatures (ISS) feature is used to extract the feature points and random sample consensus (RANSAC) algorithm is used for coarse registration with ISS feature points. Finally, the point clouds are converted into voxels with gray-value information, which will be used for DVC calculation to obtain higher accuracy point cloud registration results. Experimental results show that our method can achieve high precision registration of large point clouds and ensure sufficient registration speed.

Three-dimensional (3D) measurement based on structured light is widely applied in the diverse fields of industrial detection, face recognition, reverse engineering, and so on. Compared with typical dynamic structured light fringe projection system using DLP or DOE, the system based on MEMS mirror has the advantages of small size, low cost, no optical magnifying lens, no need for focusing, and a large field of view angle range, which can achieve sub-millimeter or even higher accuracy. In MEMS mirror-based system, structured light patterns are formed by matching the mirror angle and laser output power and accurate identification of the angle is key point for achievement of a high pattern quality. There is a big challenge in identifying the MEMS mirror’s angle since the pulse signal for driving the mirror is very narrow, around 20 ns. For meeting this requirement, a high speed and accurate FPGA is utilized to observe the narrow pulse signal of the MEMS mirror and simultaneously control the laser power output, generating precise sinusoidal fringes. A 250 MHz clock frequency for observing pulse signals position based on logic voltage comparison under an LVCMOS33 standard, successfully identifying the narrow pulse signal. This ensures a high accuracy of the light patterns. Total harmonic distortion (THD) is used to evaluate the sinusoidal property of the structured light patterns and results verify that the THD of the MEMS mirror structure light projection system controlled by FPGA is less than 5%.

Aiming at high-precision measurement of freeform surface, a method combing virtual differential confocal and reference planes monitoring is proposed. By setting the window of the CCD and selecting two acquisition areas of different sizes, the large area is used for coarse observation of the light spot when looking for the focus, and the small area is used as perform high-speed synchronous acquisition of the light intensity of the front focus and back focus. The intensity signals of the front focus and back focus are processed by normalized differential subtraction, to obtain the anti-scattering differential confocal curve. 3D point cloud data is obtained by measuring the freeform surface, and an error monitoring and compensation model is established based on standard planes to compensate 3D motion errors in point cloud data, to achieve the high precision freeform surface measurement. Theoretical analysis and experimental verification show that the axial resolution is 2 nm, the maximum measurable inclination angle of two dimensions are both 25°, and the standard deviation of measurement repeatability of 10 times PV values is 10 nm. This method can effectively expand the observation range of the differential confocal sensor, reduce the difficulty of the radial adjustment of the pinhole, and provide a new idea for high-precision measurement of freeform surface with large inclination angles and different surface roughness.

The transport sector accounted for final energy use and baseline CO2 emissions projected to approximately double by 2050 (IPCC, 2014). A solution for increasing fuel economy to significantly reduce the release of CO2 is multifaceted and needs to be scientifically and economically feasible. Fuel-borne additives is a cost effective as well as attractive method for both emission reduction and better fuel economy. As per our understanding lots of experimental studies were reported the use of mixed oxide nanoparticles for efficiency improvement but much work has not been carried out on combustion analysis in the presence of catalytic nanoparticles, which in fact is the novelty of the present work. Interferometric methods are non-contact type, which do not interfere with the field of the flame, so they cause no disturbance in the flow field and can provide full profile of temperature of flame at any instant of time. Moreover these techniques are more accurate, precise, and robust in comparison to the conventional measurement techniques. In this work Talbot interferometry has been proposed for the combustion analysis in the presence of nanoparticles due to its various advantages.

The Scheimpflug principle is commonly used in single-camera-based and multi-camera-based MFPP systems to extend the mutual overlap range of different views in the object space. We set up a dual-camera-based MFPP system and performed 3D measurements of plates, standard balls, and some specifically designed samples using the phase map stereo matching method. We conclude that the dual-camera-based system wins in measurement accuracy, while the single-camera-based system has better integrity, which may provide a reference for the system design in implementing industrial applications.

In recent years, due to the rapid development of deep learning technology in computer vision, deep learning has gradually penetrated into fringe projection profilometry (FPP) to improve the efficiency of three-dimensional (3D) shape measurement and solve the problem of phase/or depth retrieval accuracy. In order to measure dynamic scenes or high-speed events, the single-shot fringe projection technique, due to its single-frame measurement property that can completely overcome the motion-induced errors of the object, becomes one of the optimal options. In this paper, we introduce a deep learning-enabled single-shot fringe projection profilometry with a composite coding strategy. By combining an FPP physical model-based network architecture with a large dataset, we demonstrate that models generated by training an improved deep convolutional neural network can directly perform high-precision phase retrieval on a single fringe image.

The atmospheric temperature measurement uncertainty was evaluated by the Guide to the Expression of Uncertainty in Measurement (GUM) method. The lidar measurement model was introduced considering the design of actual lidar instruments and standard retrieval method. The detection noise and the auxiliary temperature uncertainty were considered as two main uncertainty sources. Based on the simulation data of Rayleigh scattering lidar operating at 532 nm with 2- hour integration period, it was calculated that two main uncertainty sources resulted temperature standard uncertainties of around 2 K and 5 K at 60 km, respectively, and the combined standard uncertainty was 6 K.

For expanding the measurement range and improvement of accuracy of multi-axes grating encoder, a mathematical model of measurement angle and diffraction spot with QPD was established. We proposed a light spot position calculation method with consideration of both the optimized composite algorithm of laser beam feature of Gaussian distribution and the QPD diagonal algorithm. In this method, we use the piecewise polynomial fitting method to fit and solved the parameters of the traditional Infinite integral algorithm and the Boltzmann function fitting algorithm. Meanwhile, we introduce a weight factor and use the Composite algorithm to compensate the spot position error. Based on the given QPD model and the basic parameters of the laser beam, simulation works are carried out and results show that the maximum error of the spot position can reduce to be an order of 10-6 mm within the 2 mm measurement range using piecewise cubic polynomial fitting, around 10% of the traditional methods.

A majority of electroluminescence spectroscopy inspection systems for LED epitaxial wafers are currently focused on conductive probe contact methods. However, the huge number of bead chips and the difficulty in performing point contact due to the microsize largely limit its application in the field of Micro-LED epitaxial wafers. In order to improve the efficiency of electroluminescence spectroscopy inspection of Micro-LED epitaxial wafers, this study proposed a practical method for contactless electroluminescence of Micro-LEDs, which can achieve contactless electroluminescence while ensuring the characteristics of nondestructive inspection. The basic element of the method is the generation of an induced electric potential by a conducting probe under a high frequency alternating electromagnetic field. The high frequency skin effect of the conductor causes a corona discharge arc at the tip of the probe. The electrical and optical properties of the contactless electroluminescence were analyzed by experiments. The results show that the error in the measured value of the spectral distribution is less than 2%. The system has an ultra-high dynamic inspection range, demonstrating the capability of efficient scanning and inspection.

For solving the problem of sub-mirror installation and posture monitoring and compensation, an absolute four-degree-of-freedom (DOF) grating encoder that is able to monitor four degrees of freedom's absolute position and pose in the θx, θy, θz, and z-direction is proposed. In this grating encoder, a grating reflector and three quadrant photodetectors (QPD) are employed and an optical path is configured based on the laser autocollimation principle. A model for the solution of the four-DOF motions from outputs of the three QPDs is established. A calibration method for the identification of the relationships between the absolute positions and QPDs outputs is proposed. A prototype four-DOF grating encoder is constructed for verification of this proposal. Test results demonstrated that the method proposed in this research can achieve absolute position distinguishing with a sub-arcsecond and sub-micrometer accuracy in rotation angles and z-direction, respectively.

Human pose estimation is a key step in understanding human behavior in images and videos. Bottom-up human pose estimation methods are difficult to predict the correct pose of a person in large scenes due to the challenge of scale variation. In this paper we propose a two-stage hierarchical network that first acquires images in large scenes, and sends tracking command signals to a two-degree-of-freedom shooting platform equipped with an image sensor to track a moving target based on a motion target detection frame, and locally constrains the captured image stream according to a top-down target detection algorithm to retain only the content related to the motion target in the image. The processed images are fed into the generalized human pose estimation model for pose detection. We deployed the algorithm on a two-degree-of-freedom filming platform equipped with camera equipment and deployed the experimental platform to sport scenes to conduct detection experiments on sport figures in running and ski jumping sport scenes, using the sport figure and its nearby area as the ROI region to generate pictures or videos with the skeleton pose of the sport target to guide the sport training of the target figure. This investigation can solve the challenge of scale variation to some extent in bottom-up multi-human pose estimation, especially for large scenes where the person key points can be located more accurately. The experiments show that this investigation can meet the practical use requirements of speed and accuracy of sport figure pose detection in large scenes of daily sports.

Optical, non-contact velocity and position measurement techniques offer great opportunities of in-process monitoring for instance in turning lathes and microfluidics. This work presents a novel laser line interferometric sensor, which enables simultaneous multipoint velocity and distance measurements. The sensor system is developed with cylindrical lens based measurement volume expansion and line camera based full field detection. Meanwhile, a matrix camera based calibration method is presented for an essential 3D evaluation of fringe spacings in the measurement volume. Experimental results show that the sensor allows synchronous multipoint velocity and distance measurements with a submicron distance uncertainty and a relative velocity uncertainty of 10⁻³ .

Digital speckle pattern interferometry (DSPI) is a full-field optical testing technique that can be used to measure tiny deformations and strains. It has been widely used in aerospace, precision manufacturing and other fields. However, the lack of effective calibration method has prevented the wider adoption of this technique. In the measurement process of DSPI, there are phase shift errors, phase noise, phase map processing algorithm errors, geometric sensitivity factors miscalibration, etc., which will lead to the final measurement error. Item-by-item calibration of the aforementioned error sources faces many difficulties in implementation and does not work well. Comprehensive calibration would be a better solution to minimize the measurement error but it is hard to perform due to the lack of suitable references for deformation measurement. In this paper, a comprehensive calibration method based on the theory of three-axis angle motions measurement using DSPI has been proposed. The tiny three-axis angle motions are loaded by Piezoelectric actuators and measured using a DSPI device based on the DSPI three-axis angle motions measurement theory. A multi-axis interferometry is used to measure the three-axis angle motions simultaneously and its output is used as the measurement reference. Because the angle motions of a rigid body instead of the deformations of an elastic body are measured, the measurement reference is readily available, yielding the successful precision calibration of the DSPI.

Novel view synthesis is a long-standing problem. Despite the rapid development of neural radiance field (nerf), in terms of rendering dynamic human body, NeRF still cannot achieve a good trade-off in precision and efficiency. In this paper, we aim at synthesizing a free-viewpoint video of an arbitrary human performers in an efficient way, only requiring a sparse number of camera views as inputs and skirting per-case fine-tuning. Recently, several works have addressed this problem by learning person-specific neural radiance fields (NeRF) to capture the appearance of a particular human. In parallel, some work proposed to use pixel-aligned features to generalize radiance fields to arbitrary new scenes and objects. Adopting these generalization approchs to human achieve reasonable rendering result. However, due to the difficulties of modeling the complex appearance of human and the dynamic sense, it is challenging to train nerf well in an efficient way. We find that the slow convergence of the human body reconstruction model is largely due to the nerf representation. In this work, we introduce a voxel grid based representation for human view synthesis, termed Voxel Grid Performer(VGP). Specifically, a sparse voxel grid is designed to represent the density and color in every space voxel, which enable better performance and less computation than conventional nerf optimization. We perform extensive experiments on both seen human performer and unseen human performer, demonstrating that our approach surpasses nerf-based methods on a wide variety of metrics. Code and data will be made available at https://github.com/fanzhongyi/vgp.

The phase-shifting algorithms are essential for a Fizeau interferometer to reconstruct the topography of the optical element’s surface or wavefront. There are differences between different algorithms for reconstruction results, especially for the suppression of noise. To acquire a more accurate Instrument transfer function (ITF) which reflects the axial spatial frequency response of a Fizeau interferometer, the algorithm transfer function which represents the characteristics of the calculation process in spatial frequency was proposed. In this paper, numerical simulations calculated and analyzed several transfer functions of the well-known phase-shifting algorithms. Then, the ITFs of a step plate with a height of 118 nm were measured with different algorithms by experiments and the results were analyzed. The simulations and experimental results indicate that the phase-shifting algorithm has an effect on the ITF measurement but it is not a key factor affecting the ITF measurement.

In this presentation, I will show how the combination of deep learning and optical measurement will bring new "vitality" to the "traditional" field of structured light 3D imaging. Compared to traditional methods, deep learning shows promising performance for applications in fringe analysis, phase unwrapping, subset correlation, and error compensation. As a result, with the aid of deep learning, we have developed a series of "single-frame" high-precision unambiguous structured light techniques for high-speed, high-precision, ultrafast 3D imaging. Finally, I will introduce our recent research on Bayesian deep learning, which promises to assess the reliability of the network by explicitly quantifying uncertainty, opening a new window for the wide acceptance of artificial intelligence in the field of optical metrology.

Scanning electron microscope have been widely studied in academia and applied in engine wear monitoring, geology, air cleanliness and pollution, because of the benefits of fast massive acquisition of nano-scale features, non-contact operation, and automatic data processing. It is important to have an automated analysis ability to get a deeper understanding on geometric features of multi-regulars shape of different particles, thus offering a significantly enhanced user experience and higher measurement accuracy. Hence, it should be carried out geometric measurement error tests before using. In this paper, several different shapes of particle were introduced to test geometric measurement error on a commercial SEM with specific particle analysis software. Several experimental cases have been designed by considered of practical application and user habits. Specially, the distinguish accuracy rate of a single threshold and multiple threshold were respectively tested by different types of particles. On the other hand, the ability of automatically classification schemes using the chemical and morphological information was taken in account too. Our scope was indeed narrow, but intentionally so. Finally, we found that these work may be of use to others who perform similar

Recently, the full-area defect inspection of high-performance optical components such as large telescope mirrors is urgently demanded. An industrial robotic arm is suitable for conducting the scanning movement of defect inspection systems, and another monitoring system is needed to guide the moving trajectory of the robotic arm. An efficient and precise guiding system is developed based on a laser projection measuring system. After the calibration of the measuring system, real-time point clouds of the component under test can be acquired. Denoising and registration of the point clouds are conducted to align the robot coordinate system with the workpiece coordinate system. Then, the scanning inspection can be conducted all over the component under test. Experimental results demonstrate that the system has high efficiency and accuracy within 17.59 μm

In order to evaluate the vertical rotating function of the rotating laser, based on the mathematical model of the vertical error, we analyzed the influencing factors of the vertical rotating function of the rotating laser, established functional evaluation parameters, built a test device for vertical error, designed an evaluation scheme of vertical rotating function. Finally, we carried out the measurement experiment of vertical rotating function parameters of the rotating laser.

A new non-contact surface roughness measurement technology has been developed. This is a technique that helps ensure the traceability of each component in the manufacturing process. Specifically, the surface roughness parameters Sq and Sdq were measured and evaluated using scattered light simulation and a light field camera. The roughness parameter Sq was determined using the generalized Harvey-Shack scattering theory and showed good agreement with existing machines. Furthermore, the roughness parameter Sdq was estimated from the argument “a” extracted from the K-correlation model.

The monitoring and measurement of in-plane coplanarity is important in the production and assembly of smaller and more fine-pitched optical and mechanical systems. Owing to its features such as greater accuracy, resolution, instantaneous response and non-intrusive nature, interferometry is advantageous and extensively utilized in a broad spectrum of applications. Interferometers possess high resolution, long measurement range and fast response times. Infra-Red diffraction interferometer is broadly used for coplanarity measurement [1-2]. Recently grating interferometer has been also used for the coplanarity measurement [3]. Circular grating Talbot interferometry has been widely used in many metrological as well as microscopic applications [4-10]. The benefits of using circular gratings are that they are easy to align, and no prior knowledge is needed for the measuring direction [10]. In this paper, a fast and cost-effective 3D (x, y, z- axis) measurement system for in-plane coplanarity inspection of an optical table is proposed. Different from the line-scanning measurement of traditional laser-based systems, the proposed, circular grating Talbot interferometer system with fringe shift measurement can provide the 3D (x, y, z -axis) measurement of coplanarity. With the conventional practice of using the zero Talbot area, it is not possible to obtain fringes of the non-coplanar surfaces unless the Talbot distance is much greater than the distance between the surfaces. The Talbot area is employed to measure the deformation of the two non-coplanar surfaces simultaneously. But this approach has a critical limitation, and the optical setup will be extremely case-sensitive. In the proposed system, Talbot interferometric fringes are extracted from the background image using image processing operations of background normalization to reduce the unevenness in the background intensity and the morphological grayscale dilation operation to eliminate the grating lines [9-10]. In the extracted Talbot interferometric fringes of these interferograms, the intensity distribution profile is plotted along the center of the grating. The center of the shifted fringe in intensity distribution profile is obtained after averaging the position of fringe. The shift between the center of two gratings determines the coplanarity. The resolution of this method is limited by the grating pitch and the capabilities of the imaging equipment.

Accurate polarization detection of light can extract a lot of optical information from them for related field analysis. A four channel detection device based on Stokes parameters is developed for dynamic polarization measurement. Four independent Stokes parameters S0, S1, S2, and S3 can be measured by this device. An energy meter is used for different incidence statuses of incident light polarization measurement. Again through the device to measure the same incident light. The polarization of the incident light is obtained by measurements of parameter calculation. Using some measurements on the polarization ratio coefficient modification, with other measurements in the revised formula calculation of the degree of polarization. The accuracy of the device is verified by comparing it with the measurement results of the power meter. By calculating Stokes parameters, the ellipse and direction angle of the single pixel polarization ellipse is obtained. The angle that was calculated can be used to obtain the polarization state of each pixel point, and draw the polarization ellipse of the whole light spot. By placing optical elements such as wave plate and polarization splitting prism in front of the optical path, the polarization state of incident light can be changed. Stokes parameters can be measured and calculated again when the light has been changed. Also, the polarization graph can be drawn. Because the change of the incident light can be controlled by the optical element, the polarization state of the incident light can be modulated. By comparing the final overall image with the polarization state of the incident light which has been modulated by the optical elements, the accuracy of the overall polarization state calculation can be verified.

Recovering high-precision 3D information of dynamic scenes from single-frame fringe pattern is a major challenge in the field of fringe projection profilometry (FPP). Inspired by the successful application of deep learning in the field of FPP, we achieve single-frame, high-precision 3D measurement through the combination of data driven and physical model-based approaches. More specifically, we utilize deep learning with powerful feature extraction ability to reduce the number of fringe images required for phase demodulation to the physical limit. Then stereo phase unwrapping (SPU) approach based on geometric constraint is used to unwrap the high frequency wrapped phases obtained from deep learning, which maximizes the efficiency of FPP without projecting additional auxiliary patterns. Experimental results demonstrate that our method can realize high-precision 3D measurement only by a single projection, overcoming the motion sensitivity problem compared to traditional methods in dynamic scenes.

Camera calibration is the bridge between 2D image and 3D world and is the basis of 3D stereo vision. Camera calibration can establish the mapping between the real 3D coordinates of space points and the corresponding 2D coordinates of the image. It will directly affect the final effect of 3D reconstruction or the positioning accuracy of 3D coordinates of space points. So accurate camera calibration technology is an essential step in 3D reconstruction, target spatial localization and even all machine visual systems, which has very important research significance. However, in some special scenarios, traditional calibration methods are difficult to use because of the long distance or difficulty in entering the scenario. To solve this problem, we propose a calibration method based on laser projection turntable. The camera shoots the laser projection lattice to determine the corresponding relationship between the pixel coordinates and the motor code amount, which is put into the calibration model to complete the camera calibration. Before camera calibration, the laser projection turntable must be calibrated with high accuracy. Firstly, the initial pose of the laser projection turntable is estimated. We accurately obtain the 3D coordinates of the calibration points with the help of the total station, and the pose information of the laser projection turntable is obtained by rigid body transformation, and then the corresponding relationship between the projection point and the code amount of the motor is obtained. Then, the least square method is used to optimize the estimation to improve the calibration accuracy. The experimental results show that the method meets the requirement of calibration accuracy and has practical application value, and the implementation of the subsequent camera calibration method can be continued.

Optical three-dimensional(3D) shape measurement technology has been widely used in industrial manufacturing, defect detection, reverse engineering, human modeling, pattern recognition and other fields. As industrial standards continue to advance, we are demanding more and more functionality and performance from our imaging systems. At present, although a real-time imaging system based on visible light from fringe projection can image well and achieve real-time imaging speed, it is still not applicable to face scanning, shaded object imaging, etc. However, most of the 3D imaging based on the infrared projector cannot achieve the effect of real-time imaging due to the slow scanning speed. In this paper, we combine the near-infrared structured light illumination system with the stereo phase unwrapping method with multi-camera calibration to realize high-precision real-time 3D imaging. The MEMS near-infrared fringe projection device is used as a structured light illumination source, which can reduce the damage of visible structured light to human eyes and animals. Experiments are carried out on static and dynamic scenes, and it is verified that the designed system can achieve high-speed and high precision 3D reconstruction at the speed of 100 frames per second, and the measurement accuracy is about 100µm.

This paper proposes a method based on the phased parametric array to solve the problem of the orientation of infrasonic waves in the air. Firstly, the principles and models of the phased parametric array are introduced, and the models are analyzed via simulation, thus proving that this orientation method can generate infrasonic waves with good directivity. At the same time, the Doyle-Chebyshev beam control method is used to suppress the sidelobes and concentrate the generated infrasonic waves in the direction of the main axis. Finally, the important parameter, namely the infrasonic wave array length, is analyzed to find out at which length the infrasound source level reaches the approximate peak. Through comparison, it is found that the actual array length is smaller than the effective array length.

In the far-filed imaging scenario, the most significant factor restricting the imaging resolution is the finite size of the aperture, which determines the diffraction limit of the system. Fourier ptychography (FP) technique provides both a wide field of view (FOV) and high resolution (HR) for imaging. Still, the efficiency is limited by the time-consuming process of image acquisition. Combined with information multiplexing technique, the imaging efficiency of muti-wavelength FP is effectively improved. However, since the same dataset of intensity measurement is used to recover the HR image at each wavelength, color reproduction problems will occur in the reconstructed image. In this paper, three additional images, captured in the R/G/B channels separately, are used as a reference to perform color calibration during the reconstruction process. Apart from that, in the subsequent reconstruction process, we use the simulated annealing algorithm and adaptive step-size strategy to correct aberrations. By updating the spectrum function and pupil function of the current sub-aperture, the HR spectrum information of the measured target is obtained. Our approach is verified by simulations where the amplitude RMSE values of R/G/B channels are less than 0.071, 0.076, and 0.064 respectively. The resolution of objects is improved approximately fivefold, which is consistent with the theoretical reconstruction multiple.

Speckle projection profilometry (SPP), as an efficient 3D measurement method based on structured light projection, projects the speckle pattern based on spatial encoding onto the measured scene to enhance its texture, thereby improving the accuracy of single-shot 3D measurement. However, the traditional stereo matching method in SPP compromises the measurement accuracy in order to ensure the robustness of 3D measurement. At present, some speckle matching methods based on deep learning have been proposed to obtain high-precision and dense disparity maps, but at the cost of expensive computational overhead, which occupies a lot of memory resources and reduces the running speed. Different from existing networks, this paper proposes a lightweight end-to-end stereo matching network by combining attention mechanism, spatial pyramid pooling module (SPPM), and multi-scale feature fusion, which achieves single-shot 3D measurement with competitive accuracy while running at 170 ms.

Deep learning is currently gaining a lot of attention in the field of optical metrology and has shown great potential in solving various optical metrology tasks such as fringe analysis, phase unwrapping, and hologram reconstruction. For fringe analysis, current major works use U-Net and its derivatives as the backbone of the deep learning network, but suffer from a large number of model parameters and computational redundancy of the U-Net network, which outputs low-precision prediction results while taking up a lot of GPU memory. To solve these problems, compared with U-Net, a lightweight fringe analysis network with the size of only 1.7G is proposed to reduce the memory usage by over 70%, while improving the accuracy of phase retrieval by 10%, providing a new path for the widespread implementation in mobile devices of deep learning-based optical metrology.

The 3D measurement technology based on speckle projection has been widely used in emerging fields such as intelligent processing and manufacturing, face recognition. because of its advantages of noncontact and full-field measurement. In this paper, we develop a high-precision 3D sensor system based on multiple infrared speckle projection modules to obtain highly detailed 3D reconstruction data by projecting speckle patterns at different angles. In this system, we design the speckle projection module to encode the depth information of the measured scene by adjusting the laser angle in real time, and then, combine a coarse to fine spatial-temporal stereo matching strategy to improve the accuracy of 3D measurement. Finally, in the 3D measurement experiments of complex multi-object scenes under a large field of view, we verify that the actual measurement results of our system have high-completeness.

In stereo vision, depth information is obtained by establishing a spatial correspondence between the two cameras based on the triangulation, so it is important to maintain high-accuracy calibration parameters of the stereo cameras. However, under some extreme conditions, such as high temperature, high pressure, and strong vibration, there are irreversible changes in the parameters of the camera lens and the spacing between two cameras, which leads to the invalidation of known calibration parameters. In this paper, a stereo camera self-calibration method is proposed to get high-accuracy feature point pairs from speckle planar image pairs using the DIC-based grid point matching technique. The calibration result of cameras is corrected after SVD and RANSC iteration, which can enhance the quality of stereo rectification and the 3D measurement accuracy of stereo vision.

During the process of automobile manufacturing and transportation, it is inevitable to cause automobile surface defects, such as scratches, sunken, blots, and so on. This will seriously affect auto sales and lead to huge economic disputes between transportation companies and consumers. At present, manual detection is still the mainstream way of defect detection for automobile surfaces, which is unstable and time-consuming. This paper presents a defect detection method for automobile surfaces based on a lighting system with light fields. Fast, automated, and accurate location of surface defects can be achieved by using a high-quality defect imaging method based on light fields, the multi-exposure fusion algorithm, and the YOLO V5 network. For different materials or surfaces reflection characteristics, the proposed method can accurately detect various surface defects in areas such as doors, windshields, and wheel hubs.

The LED market in horticultural lighting is growing rapidly, and there is an increasing demand for horticultural LED calibration. The evaluation of LED lighting for plant growth needs to be based on photon systems, such as photon flux (PF) rather than a photometry system which is used to evaluate how bright a LED is for human eyes. The integrating sphere and spectroradiometer are selected as the key instrument of the test system for LED photon flux calibration. Standard and improved calibration methods are presented. The standard method includes calibration of the test system, verification, spectral absorption correction, and measurement of the test LED lamp. The improved method simplifies the absorption correction in the calculation. Three LED lamps for plant growth are calibrated by the two methods, and the photon flux scale is traced to total spectral radiant flux (TSRF) standard lamps with the relative uncertainty Urel = 2.5% (k = 2).

An illumination system for coherent noise suppression is envisioned. The principle of coherent noise suppression is introduced, and this illumination system is designed this way. The illumination system is simulated, and the hybrid sequence model of Zemax analyzes the irradiance. To verify the noise suppression effect of this illumination system, it is used as the illumination part of the Fizeau interferometer. The interference process is simulated under the nonsequential model of Zemax to obtain an off-axis light source with a ring radius of 0.32 mm and a numerical aperture of 0.14. The Fizeau interferometer with a conventional light source is also simulated. A comparison experiment is set up to generate the same noise point in the two interferometers using different illumination modes to trace and produce the same four interference fringes with the same interferometric cavity length of 20 mm and the same tilt angle of the measured surface of 0.057°. Compared with the interferograms in the conventional illumination mode, the interference fringes formed by the illumination of this study are almost undamaged, and the near-complete interference information can be retained. The interferometer system with this light source was built and the test results were verified, and it was found that it could achieve the measurement accuracy of 1/20 wavelength and the measurement stability of 1.219 nm, and it also had a good contrast of interference fringe.

Wafer, the primary material used to make semiconductor chips, are found in almost every type of electronic device used in everyday life. As the quality of wafer used in large-scale integrated circuits has improved considerably, the diameter of wafer has continued to increase, and the thickness of silicon wafer has become increasingly thin. Wafer manufacturers and device manufacturers are increasingly focusing on wafer thickness variation. In the past few years, the usual capacitive tools for wafer inspection have been replaced by interferometric tools for higher sensitivity and resolution. We, therefore, describe a method that uses two Fizeau-type phase-shift interferometers to simultaneously measure the front and back surfaces of a vertically placed wafer and calculate the thickness variation of the wafer based on the resulting morphologies. The reliability of the method was verified by comparing the wafer thickness variation obtained from experimental measurements with that obtained from optical glass bonding. Over three days, five consecutive measurements were performed daily on 50mm wafer using this method, and the experimental results showed that the average values of RMS (Root Mean Square) of the thickness variation calculated for each day were 41.843nm, 40.751nm, and 40.490nm, and the average values of PV (Peak to Veally) were and 206.761nm, 205.252nm, and 209.800nm, and the measurements proved to be highly reproducible. The method has good stability and reliability to meet the measurement of wafer thickness variation.

The laser displacement sensor based on the triangulation method has the advantages of large range, high precision, and strong anti-interference. Due to the limitation of the measurement principle, although the laser displacement sensor has high repeated measurement accuracy, it has serious high-order nonlinear system errors. Considering that in some measurement occasions, only a few local measurement intervals within the range are used. In this paper, a local nonlinear error calibration algorithm based on C2 continuous interpolation was proposed. A nonlinear calibration model was established by intensive calibration data acquisition in the local interval of the measurement application. The model can be smoothly connected with the original linear calibration model. At the same time, the measurement accuracy of the local area is greatly improved without affecting the accuracy of other intervals.

Photomask manufacturing is an essential cornerstone of the semiconductor industry. Photomask substrate made of quartz is a hot topic in photomask manufacturing today, and the high requirements of photomask on quartz substrate make its optical homogeneity index extremely important as well. When measuring the optical homogeneity of a quartz substrate using a four-step absolute measurement method based on the Fizeau phase shift interference principle, the phenomenon of overlapping interference fringes was discovered. The cause was analyzed and the structure of measurement was simulated by Zemax. The simulation results identify the non-parallelity of the front and rear surfaces of the quartz substrate as the main effect.

To address the problem of high-precision detection of large flat aperture mirror shapes, the current subaperture stitching methods are analyzed. The results show that the full aperture surface shape stitching errors due to subaperture adjustment errors and localization errors are the main reasons affecting the high precision detection. To improve the detection accuracy of subaperture stitching, we propose an immune optimization algorithm for subaperture stitching. The algorithm reduces the influence of subaperture adjustment errors and positioning errors, so that the PV value of reconstructed shape reaches λ/100. It is shown that the proposed method can effectively control the stitching errors and improve the stitching accuracy.

Ring elements are of great significance in the field of precision machining, and their performance is crucial for the functionality of the whole system. However, the present frequent inspection of such components lacks effective means and is prone to causing damage to the component. In view of this, this article on the basis of comprehensive research, combined with the principle of optical interference detection, puts forward a measuring method for such components, the method by introducing a pyramid to realize high precision optical path, and according to the inner surface in the shape of different components, it can be divided into cylinder and circular arc surfaces, respectively, measuring scheme is designed. Then, according to the designed inspection scheme, the system is built separately for inspection. After the system is stable, the shape information of the tested part is obtained. After the system is stabilized, the shape information of the inspected part is obtained. The shape information of the inner surface of the ring element can be obtained after only one inspection.

The traditional interferometry method of large aperture optical elements relies on changing the beam expansion lens and optical path structure artificially according to different test samples, which inevitably introduces some systematic errors. Therefore, this paper proposes a corresponding dual-wire control scheme according to the functional requirements and simulation experiments of dual-optical path interferometer. This scheme is based on serial communication protocol and Zigbee communication protocol. Through the coordination of Zigbee wireless control, serial software control and mechanical structure, the optical path can be folded and calibrated for many times, so that the changing position of the optical element after each switch of measurement aperture is fixed. The real-time status is displayed in the interactive interface developed based on MFC (Microsoft Foundation Classes). Finally, the whole system is tested and verified. The results show that the system can basically complete the real-time debugging of the optical path, which provides a practical design idea for the automatic control of the hybrid interferometer in the future.

Object detection and tracking in football video is a very challenging task, and it has good practical and commercial value. The traditional method of extracting the target movement trajectory of football matches is often carried out by players carrying recording chips, which is expensive and difficult to popularize in amateur stadiums. There are also some studies that only use the camera to process the targets in the football video, but due to the similar appearance and frequent occlusion of the targets in the football video, these methods can only segment the players and the ball in the image, but cannot. Track it or only for a short period of time. We study active object tracking method for football game, where a tracker takes visual observations (i.e., frame sequences) as input and produces the corresponding camera control signals as output (e.g., turn up, turn left, etc.). Conventional methods tackle tracking and camera control tasks separately, and the resulting system is difficult to tune jointly. These methods also require significant human efforts for image labeling and expensive trial-and-error system tuning in the real world. To address these issues, we propose, in this paper, an end-to-end solution via deep reinforcement learning. By building a football game simulation scene in the simulator (Unreal Engine), the entire field can be covered by turning the camera in the simulation scene.

3D human pose estimation refers to recovering the pose of the human body in the three-dimensional space from the image, which is estimating the coordinates of the key points of the human body in the three-dimensional coordinate system. The human skeleton model composed of these keypoints describes the human pose in the image.The traditional pose estimation method uses motion capture equipment, which requires athletes to wear auxiliary equipment and affects the normal training of athletes. With the development of deep learning, convolutional neural network has shown strong representational capabilities in the process of image feature extraction, and is an effective method to achieve human pose estimation. However, in the actual complex and changeable three-dimensional space, the environment is noisy and the limbs are occluded. The monocular capture of human posture has the problems of low accuracy and instability, and cannot provide sufficient semantic information. To solve this poroblem, this paper designs a multi-view information fusion algorithm. First, a two-dimensional attitude feature extraction module is designed. By connecting and interacting heat maps of different resolving power in parallel, the heat map always maintains a higher resolving power for higher precision. It extracts the two-dimensional human pose from a single perspective, and then uses the multi-view information fusion technology proposed in this paper to complete and correct the pose information from different perspectives to obtain a more accurate two-dimensional pose. Then, through the camera pose transformation, the three-dimensional human pose estimation is realized, and the experimental verification is carried out on the multi-eye machine vision pan-tilt zoom tracking shooting system.

We adopt Unreal Engine to construct a virtual football stadium. In the virtual football stadium, we set two teams of AI players controlled by behavior trees, and simulate football games by playing against each other. At the same time, we deploy a virtual camera in the auditorium that can be rotated horizontally and vertically to capture the picture. We adopt the Transmission Control Protocol communication plugin in the mall to realize the communication between Unreal Engine and programs. Saving the current frame picture locally through the built-in screenshot command of Unreal Engine, and send the internal game information to be read to the client at the same time. The client reads the local image, accepts the information and sends the corresponding action to the Unreal Engine. Unreal Engine executes the received action to rotate the camera.

In the process of measuring the Fizeau interferometer, the measured plane mirror is inevitably affected by gravity, which leads to the general situation that the actual plane shape is inconsistent with the measured result. The study shows that the effect of gravity on surface deformation is negligible for small apertures (about less than 100mm), but it is especially obvious for the precision measurement of large-aperture (more than 300mm). In order to reduce the influence of gravity deformation on the measurement results, this paper simulates a variety of fixing schemes for 600mm plane mirrors, selects a better large-aperture plane mirror fixing scheme finally, and obtains the simulation results of 600mm plane mirror surface displacement PV value is 10.79nm, which is less than 1/50λ (λ=632.8nm). It provides a theoretical basis for follow-up experiments.

Nowadays, linear displacement and angle measuring devices are widely used in the assembly, calibration and deformation monitoring of industrial structures. The Optic-electronic autocollimators for non-contact measurements are highly effective in these cases, since the measurement has a large measuring distance and high accuracy, but the existing autocollimation sensor for measuring angular and linear coordinates consists of two separate measuring systems and two reflectors, respectively. This disadvantage is a major obstacle for determining 6 motion coordinates at a point of the object. Autocollimator is suggested, capable of determining 6 object motion parameters (3 3 rotation angles and 3 linear displacements). The two main components of the autocollimator are the special reflector at the track point of the monitored object and radiating-receiver unit at the rigid base. A specially constructed tetrahedral prism is used as a reflector. Two emission marks are placed directly in front of the radiation receiver objective. Each emission mark after reflecting through the faces of the reflector produces 6 images, so with two emission marks 12 images are produced behind the reflector. The displacement of the object is calculated from the coordinates of the images obtained on the sensor. Mathematical models that determine the displacement of the object based on the image coordinates on the matrix photo receiver have been built. A comparative analysis between the optic-electronic autocollimators and the existing schemes was performed by computer simulation. The simulation results show that the proposed scheme has a smaller error. In addition, the simple and compact structure is also one of their outstanding advantages.

With the progress of technology, 3D printing technology has gradually developed and matured, and has been widely used in aerospace, game artifacts, medical industry and heritage conservation. As a reverse engineering technology of 3D printing, 3D reconstruction technology can reconstruct a 3D mesh model from several images of typical target entities. Monocular 3D reconstruction has the advantages of easy access and low cost. Therefore, this paper improves a multi-view stereo reconstruction technique based on monocular camera: DC-MVSNet. Firstly, the collected multi-view images are sparse reconstructed by colmap to obtain sparse point clouds and camera poses. Then, it is input into DC-MVSNet network and output to get the depth map corresponding to the reference image. Finally, the 3D point cloud model is obtained by deep fusion. In the feature extraction module and depth map refinement module, a densenet and a coordinate attention mechanism are added respectively to improve the feature extraction ability. The proposed method is compared with previous works. The results show that the reconstruction completeness of the algorithm for weakly textured objects is improved in both objective quantitative indexes and subjective perception. The study can be deployed on win10, linux and embedded systems, working reliably and practically. It is of significant reference value for 3D printing reverse engineering.

Deflectometry is a slope-based technique to measure specular surfaces. Modal reconstruction methods fit the surface shape with a certain mathematical model based on expansion polynomials and their coefficients. The coefficients are obtained by linear equations, which are consisted of the gradient of the polynomials and the measured slope data. Nevertheless, computing the large linear equations is time-consuming work and the noises and outliers will decrease the reconstruction accuracy. This paper uses the Chebyshev polynomials as the basis set and proposes a modal reconstruction method based on the deep convolutional neural network to directly output the corresponding Chebyshev coefficients. Compared with the conventional modal reconstruction method, the results demonstrate that the reconstruction accuracy and the computational efficiency are improved effectively using the proposed method.

Moving target trajectory prediction is a typical multidisciplinary research issue involving intelligent science and technology and transportation engineering. Automatic tracking and shooting is a field with significant theoretical research and practical application values in the problem of moving target trajectory prediction. With the increasing demand for tracking shooting, the requirements for target tracking are also getting higher and higher, which has attracted widespread attention in recent years. In the last few years, automatic shooting has gained attention. It can automatically follow athletes and event training to provide more competition videos and also serve as a guiding tool for competition monitoring. Due to the complexity of the target trajectory, tracking shooting has gradually become one of the difficulties in the researches of the area. Most of the traditional methods of trajectory prediction use mathematical models to predict the states and behaviors of targets, however, the computational complexity of traditional methods is high and they show poor performance in real scenarios. The tracking shooting system proposed in this study converts the detected three-dimensional moving objects into two-dimensional images by joining the turntable, employing encoder-decoder model and using an end-to-end computing method based on long-term memory network, and makes full use of the position and action posture information of acquired moving objects, so that it can deeply mine the historical position information and behavioral habit semantics of moving targets and realize effective mapping with future trajectories, predict the position information at the future moment, and then control the turntable to realize the tracking and shooting of 3D moving targets.

Heterodyne grating interferometer is widely used in precision positioning due to its high precision and robustness. However, the polarization states of two frequency components in a dual-frequency laser are easy to overlap with each other because of the non-ideality of optical components. It will cause nonlinear error, which limits the measurement precision of the grating interferometer. To improve the frequency aliasing of heterodyne grating interferometer, a polarization adjustment module is proposed to adjust the polarization angle of the dual-frequency laser. The dual frequency system outputs two laser beams with different frequencies separately. The module realizes the polarization adjustment of the two frequency components through two groups of the polarizers and the half-wave plates (HWP). Finally, the polarization directions of two frequency components are orthogonal and combined by the beam splitter (BS). Thus, the nonlinear error caused by frequency aliasing is removed. The polarization adjustment module has the advantages of not changing the direction of the laser propagation and simple structure, which makes it easy to realize integration. It can provide a reference for the solution of frequency aliasing of heterodyne grating interferometer.

Aiming at the low accuracy of behavior recognition technology for multi-target human behavior recognition in small and medium scenes, a method for multi-target human behavior recognition in small and medium scenes is proposed. In this paper, YOLOv5 and DeepSort are used to detect, track and locate human targets in the video stream. According to the detection frame, the appropriate size of the human target is cropped as the input image of the behavior recognition module to reduce the interference of human behavior background, and finally realize the multi-target human body behavior recognition. The behavior recognition module is composed of an improved C3D network, and the features extracted by YOLOv5 are shared with the behavior recognition module to reduce the amount of computation. Experiments show that this method achieves end-to-end recognition,and can recognize the behavior of different target human bodies in small and medium scenes, and achieves comparable results.

Ophthalmic optical equipment is a kind of clinical medical equipment based on optical principles, which can be used for ophthalmic examinations and diagnoses. Their imaging qualities and measurement accuracies are very important. Therefore, regular calibrations with ophthalmic optical standard models are needed. Aiming at the metrology of axial parameters of ophthalmic optical standard models, a time-domain OCT metrology system based on multi-spectrum is developed. It can measure and calibrate the optical path values or physical thicknesses (refractive indexes are known) of each layer inside the models at different wavebands. The developed metrology system has the advantages of high speed, non-contact, disassembly free, and application to complex structures. To meet the metrology needs of ophthalmic optical instruments in different working bands, three commonly used bands can be switched in this system by an optical switch, whose central wavelengths are 840nm, 1060nm, and 1310nm respectively, with a bandwidth of around 60nm. The reference arm of the system adopts a high-precision electronic control optical path matching device, and the sample arm can achieve a maximum detection depth of 80mm in air. Combining dispersion compensation mirror and algorithm, a clear interface edge and an axial resolution of 10μm are obtained. In addition, the axial length standard model recommended by ISO 22665 and the anterior segment standard model developed by the national institute of metrology were measured experimentally. In the experiment, refractive indexes of materials in each waveband are known, and the maximum error of measuring each layer’s thickness is less than 10μm. In a word, the developed system in this paper can effectively solve the axial parameters’ metrology and calibration problem of ophthalmic optical standard models, especially with large depth and complex internal structures.

Recently, the method of multipoint interferometer (MI) was proposed to detect the orbital angular momentum (OAM), which could be used for measuring OAM of light from astronomical sources. However, this method is limit to low topological charge, because the interference pattern can repeat periodically. In order to solve this problem, we proposed an improved multipoint interferometer method (IMI) capable of measuring the OAM of optical vortex with arbitrary topological charge. However, the method of IMI is limited by complex interference patterns detection difficult. To overcome these disadvantages of MI and IMI, we proposed a method called DRI using screen plate with a binary ring slit and two pinholes. This method is demonstrated viability theoretically

In order to solve the problem that the field calibration technology of pulsed laser micro-energy is not complete, which makes it difficult to calibrate the pulse energy parameters of Femto-Joule level pulsed laser source output, the field calibration device of Femto-Joule level laser energy is developed by using the principle of analog integral. By establishing the mathematical model and numerical simulation analysis of the photoelectric detection module, the device optimizes the cutoff frequency of the photoelectric detection module, reduces the minimum detectable power of the photoelectric detection module to 6.7nW, improves the micro-energy laser detection capability, reduces the transmission bandwidth and complexity, and improves the equipment integration and anti-interference ability to meet the requirements of field measurement and calibration. At the same time, the device adopts the integral circuit structure based on the transconductance amplifier, analyzes the relationship between the integral window time and the integral error through the numerical simulation method, accurately controls the integral window time, reduces the integral error to 4%, effectively suppresses the stray light interference, and improves the field measurement accuracy of Femto-Joule level pulsed laser energy measurement is improved. In order to comprehensively examine the performance of the field calibration device of Femto-Joule level laser energy, an experimental verification device is designed, and the verification and calibration tests are carried out. The test results show that the field calibration device of Femto-Joule level laser energy can accurately measure the laser energy value of pulse with 1ns~100ns pulse width. The energy measurement range of the device is 10fJ~1pJ, and the measurement uncertainty is 14% (k=2), The technical parameters of the device can meet the requirements of field calibration of Femto-Joule level laser energy.

An inline fiber Mach-Zehnder Interferometer (MZI) that consisted of a standard single-mode fiber sandwiched between one peanut-shape and a core-offset joints is fabricated by a fusion splicer for RI sensing purpose. After substituted a peanut shape for a core-offset in an inline fiber MZI with the core-offset pair, the influence caused by the misplaced direction of the core-offset is avoided since the peanut structure is symmetrical, and it is found that its RI sensing performance was improved as well. The inline fiber MZIs with the different arm lengths and core-offset displacement were evaluated for refractive index (RI) measurement. The RI sensitivity of -72.4 nm/RIU was achieved within the measured RI range of 1.333–1.373.

The Sun is one of the most vital celestial bodies within solar system. Scientific researches from China on the Star Sun are impressively emerging in the recent decade.At present,China has put forward a mission similar to Parker Mission。 As one of the tasks supporting this mission, the space environments within a few solar radii have been simulated and assessed in this paper. It is necessary to make certain what severe situations the spacecrafts should be faced up to before the mission is granted. Simulations have shown the radiation level for the near Sun orbiters at altitude of several solar radii is much more severe than that for the near Earth orbiters at least by 1000 times. 20 millimeters of aluminum shielding might effectively reduce the radiation level of the spacecrafts.Hundreds of Krad (si) radiation will be accumulated in a five-year mission even though 20 millimeters of aluminum shielding is adopted to relieve the radiation effects.

The article presumes a data processing algorithm that improves the accuracy of recognition of radio-electronic components in devices for automated installation. The paper proposes the use of a multicriteria filtering method that allows you to automatically change the smoothing coefficient. Varying the coefficient allows both reducing the noise component and preserving the boundaries of the radio elements without blurring. In order to enhance the contours of objects, a data simplification method is applied using the technique of reducing the range of clusters of color gradient histograms while preserving the shapes of objects. At the stage of detecting the boundaries of the elements and forming the structure of the elements of the radio component base, a modified one-dimensional two-criteria method is used. The combined analytical approach allows detection of the boundaries of radioelements and increases the productivity of the process. As test data used to evaluate the effectiveness, pairs of test images obtained by sensors fixed at various magnifications with a resolution of 1024x768 (8 bit, color image, visible range) are used. Images of simple shapes are used as analyzed objects

The article develops an approach to automated identification of the accuracy requirements set in the detail drawing. A technique for recognizing accuracy requirements based on image analysis is proposed. The algorithm for identifying tolerances on linear sizes is based on classical text recognition algorithms. The advantage of the developed approach is its versatility. The effectiveness of recognizing tolerances on linear sizes does not depend on the options for setting and orientation of text entries in the drawing. A database of tolerances on linear sizes has been developed, which makes it possible to increase the efficiency of identifying accuracy requirements by comparing recognition results with standard values. The structure of a convolutional neural network for identify the symbols of tolerances of form, orientation, location and run-out, roughness, is proposed. This makes it possible to determine with high accuracy the area of requirements and improve identification performance

The article developed an approach to improve the accuracy of measuring deviations from straightness and flatness of extended surfaces. One of the most accurate and productive methods for measuring deviations from straightness and flatness of extended surfaces is the step method using electronic levels. The main difficulty of the measurement lies in the correct positioning of the electronic level on the measured surface. Algorithms for processing the results of measurements assume the coincidence of the measured points during successive measurements. Otherwise, a significant methodological error is introduced. A technique for recognizing the position of the level on the measured surface is proposed based on the analysis of images obtained from video cameras. The advantage of the algorithm is its resistance to determining the position of the level, regardless of its orientation on the measured surface. Software has been developed to correct the position of the level on the measured surface. An algorithm for determining the adjacent plane based on the gradient descent method is proposed. The dependences of the methodological measurement error on the number of measured points and the accuracy of their determination are revealed.

This article presents a two-stage approach, combining novel and traditional algorithms, to image segmentation and defect detection. The first stage is a new method for segmenting fabric images is based on Hamiltonian quaternions and the associative algebra and the active contour model with anisotropic gradient. To solve the problem of loss of important information about color, saturation, and other important information associated color, we use the quaternion framework to represent a color image to consider all three channels simultaneously when segmenting the RGB image. In the second stage, our crack and damage detection method are based on a convolutional autoencoder (U-Net) and deep feature fusion network (DFFN-Net). This solution allows localizing defects with higher accuracy compared to traditional methods of machine learning and modern methods of deep learning. All experiments were carried out using a public database with examples of damage to the TILDA fabric dataset.

A new approach to the automated design and control of ceramic end mills is proposed, which allows creating a group of structures for machining a range of products from various hard-to-cut materials in different modes. The method compares favorably with the existing ones by creating new cutting tool with increased performance, providing an increase in the resistance of mill to brittle fracture up to 2-3 times. The design approach includes a comprehensive measurement module for quality control of products with a reduction in the complexity of measurement up to 10 times in terms of time.

In this paper, a set of indicators for an effective assessment of the cutting ability of a grinding wheel was established based on the detection of the degree of filling and immersion of abrasive edges. It has been pointed out that a profile of a populated circle has a direct correlation with the average number of recognized faces. A new indicators for measuring zones with active edges and their distribution on the surface of a grinding wheel was developed, which can be identified by the models algorithms, is formed

A method for microprocessing products with a shaped generatrix by remote the product from the image is proposed, which provides an increase in productivity without loss of quality. The method allows you to recreate an object based on image reconstruction with basic accuracy requirements and establish a rational trajectory of the turning tool on CNC machines. The method is implemented as follows: a cylindrical workpiece is fixed in the machine spindle, the plate is installed in the turning body cutter, a preliminary positioning of the cutter is performed and its fixation in the working area of the machine with a special trajectory of movement obtained on the basis of the recognized profile of the product during reverse engineering. As a result, the new method allows increasing the productivity of the treated surface up to 2 times, depending on the shape and accuracy of the object being reconstructed.

In mechanical engineering, cutting tool wear is a major factor in the functional and parametric failure of a cutting tool. The loss of tool performance causes failures in the production of suitable products, as well as machine downtime. In this connection, accurate measurement of wear of the cutting wedge at various stages of the technological process is extremely important. The article proposes a method for recognizing wear zones on the back surfaces of cutting inserts by image processing. Indicators have been formed for assigning a failure criterion for various quality factors. For the formation of a measurement system, a three-stage control method with the rationale for operational recommendations is proposed. The correlation of wear by image processing has been compared with the optical method currently applied in manual mode, which greatly reduces the efficiency of production quality control.

The geometric parameters of the flute profile of micro-mills have a great influence on both the strength characteristics of the tool and the cutting process. This explains the need for high-precision control of the shape of the flute profile. In modern CNC controlling and measuring machines, a non-contact measurement technique is implemented using high precision reflected light cameras operating based on the principle of contrast autofocus. The existing methods of recognition of the flute profile have large errors associated with the low efficiency of the contrast focusing method in the case of control of a section of several subsections. The proposed recognition algorithm is based on the search for local focus points obtained as a result of the analysis of images from the reflected light camera. This method is shown to provide high accuracy control of the flute profile of micro-mills.

Cutting tools with shaped cutting surfaces are widely used in industry. Their use allows to improve cutting conditions and to increase productivity. The faceted cutting surface is formed by cutting edges of complex shape. Since, there does not exist a standard/conventional methods for controlling a specialized tool, the development of methods for controlling chips of a cutting wedge of a complex shape is of high importance. In this paper, an image processing algorithm that allows to search for the shape of the edge and find deviations in the form of a chip from the theoretical result has been developed. The algorithm helps to quickly transform from a digital image to the real scale of the cutting edges. The translation of the coordinate system of the measuring machine calibrate and control the angular position of the cutter during measurement. The approach to finding the coordinates of the cutting edge, developed in this article, includes an error elimination block with experimental verification of functional dependencies, which allows to quickly correct the angular position of the controlled tooth online. This can significantly increase the efficiency and speed of measuring end mills for adjusting the angular position and correction when controlling a multi-blade tool while evaluating the accuracy of the cutting edge and the magnitude of the chip on the back surface.