An automatic spot location and extraction algorithm based on the combination of the Otsu method and the centroid method is proposed herein for the grating wavefront curvature sensor and verified through experiments. The original intensity distribution image is first binarized by using Otsu method. Then, the binary image is then segmented into two binary images according to the centroid coordinates such that each binary image contains one spot. Finally, the centroid coordinates of the two images are calculated as two spot centers in the original intensity image, while the two spots in the original image are extracted from the original intensity image. An experimental setup of the grating wavefront curvature sensor is built based on an off-axis Fresnel zone plate to verify the effectiveness of the proposed algorithm. In the experiment, the spots are extracted automatically by using the above algorithm; then, the wavefront is restored by using Laplacian eigenfunctions. The results are compared with those of the Hartmann wavefront sensor. The experimental results demonstrate that the proposed algorithm can extract spots automatically with an error less than 4 pixel.

An adaptive bit loading algorithm is proposed based on visible-light pixelated multiple-input and multiple-output (MIMO) systems to deal with the problem of high number of bit errors for some subcarriers caused by the high-frequency attenuation, which is attributed to the absorption and scattering of channels and lens blur in underwater visible-light pixelated MIMO systems. By ensuring that the total number of bits allocated across all the subcarriers remains constant, the proposed algorithm dynamically selects the optimal subcarrier modulation order according to the signal-to-noise ratios of the subcarriers to reduce the number of bit errors for each subcarrier and improve the system reliability. The simulation results denote that the bit error rate can be considerably reduced when the transmission rate of the pixelated MIMO system is maintained constant.

Herein, a bilayer sub-wavelength metallic grating structure is optimized in the visible-light band using genetic algorithm and finite-difference time-domain method to improve the grating performance. Through simulation analysis, the structure can achieve polarization filtering function of blue band with a maximum transmittance of 77%, maximum extinction ratio of 20 dB, and full width at half maximum of 35 nm. By analyzing the influences of grating parameters on grating performance, it is proven that the structure has a large process tolerance. In comparison with the conventional structure, the optimized grating has good transmission performance and filtering performance, and it has good application prospects in fields such as polarization navigation, polarization detection, display, and imaging sensing.

To ensure good communication quality of several users from different orientations within a room in indoor visible light communication systems, this study proposes a power optimization method based on the improved ant colony algorithm, which converts the one-dimensional path optimization problem into three dimensions and improves the operational efficiency, thereby improving the uniformity of the received optical power distribution of the receiving plane. Further, the difference between the maximum and minimum powers of the receiving plane is considered to be the optimization target. The addition of constraints avoids the occurrence of low average optical power after optimization as well as the uncertainty attributed to manual screening. Finally, the simulation results demonstrate that the proposed method reduces the optical power offset and ensures the stability of the communication environment.

The traditional spectral matching method is based on the spectral reflectivity. However, the process of obtaining the reflectivity of each point is complicated, and the calculation has some errors, which will affect the recognition accuracy. In order to solve this issue, the problem of mural pigment recognition has been transformed into multi-spectral image classification, and a convolutional neural network algorithm with strong advantages is used in image classification to process multi-spectral images. Meanwhile, a new convolution neural network model is designed, and a data preprocessing method of spectral feature reorganization is proposed. By adding two dropouts, the problem of over-fitting in the training process is prevented, and the classification of ancient mural pigments is realized. The experimental results show that compared with the statistical manifold support vector machine classification method and the convolutional neural network classification method without dropout, the proposed method has obvious advantages in classification effect and classification accuracy.

The loss function of a target pose estimation model based on a convolutional neural network (CNN) mostly uses the Euclidean distance between two points as the evaluation criterion. Although the loss function is simple in calculation and fast in operation, the training rules are not comprehensive enough and lack global understanding of the target. In this paper, a ComPoseNet model based on a combined loss function is proposed for pose estimation. The loss function in this model is based on spatial learning, and the two-point Euclidean distance, straight line, and straight line angle are used as training rules. Compared with the traditional loss function, this algorithm considers the spatial position of the target from the point, line, and angle, reducing the error between the estimated and the real poses so that the effect of the pose estimation is improved. Numerous experiments and analysis of LineMod data show that the algorithm has a higher convergence speed, greater accuracy, and smaller errors than the traditional algorithm operating for the same training times. The translation error is reduced by 7.407%, and the angle error is reduced by 6.968%.

To solve high computational resource requirement for running hardware platform of the series of the YOLO object detection method due to the huge parameters, the large amount of calculation, and the large scale of detection model, this paper developed a light-weight object detection network based on inverted residual structure(IR-YOLO). First, it used depth separable convolution to reduce detection model parameters and computational quantities. Secondly, it constructed inverted residual block based on depth separable convolution to extract high-dimensional feature. Finally, according to the characteristic of inverted residual structure, it used a linear activation function to reduce the information loss during the process of channels combination. The experimental results show that the IR-YOLO detection model is reduced by 47.7% compared to the YOLOv3-Tiny detection model, it validated that the IR-YOLO algorithm can effectively compress the model while maintaining detection accuracy.

This study proposes a super-resolution reconstruction algorithm with hierarchical clustering dictionary training and similar constraints for astronomical images, according to the theory of centralized sparse representation based image super-resolution reconstruction, thereby solving the problem of low imaging resolution of the astronomical images. In the dictionary training phase, a novel hierarchical clustering algorithm is used for classifying the sample image patches. Further, each image patch is independently trained to obtain multiple compact dictionaries. In the image reconstruction stage, the accuracy of the sparse coding coefficients is improved by suppressing the sparse coding noise. Subsequently, the sparse coefficients of the reconstructed image can be reasonably estimated based on the non-local self-similarity of the image. In addition, the image reconstruction process is globally constrained by the construction of non-local self-similar regularization terms. The experimental results denote that the proposed algorithm can effectively improve the resolution of astronomical images. Furthermore, the subjective visual effects and objective evaluation indicators of the reconstructed images are observed to be superior to those obtained by using other traditional super-resolution reconstruction algorithms.

A new calibration method is proposed to address the limitations of the line-structured light calibration method in the existing measurement systems. The proposed method improves the extraction accuracy of feature points by avoiding intersection between the light stripe and chessboard on the target reference,and improves the calibration accuracy by increasing the number of feature points for light plane fitting. The coordinates of feature points in the camera coordinate system are computed by combining the equation of rays which is formed by the connection of arbitrary feature points at the center of the light strip and the camera center and the plane equation of the target reference, then the coordinates of the center points of two and more strips in the camera coordinate system are solved. Finally, based on the least square method, the equation of the light plane in the camera coordinate system is obtained based on coordinates of all center feature points of the stripes. Experimental results show that the accuracy and robustness of the proposed method are superior to those of existing calibration methods.

This study presents an effective method for segmenting the skeleton of a three-dimensional point cloud. Segmentation results can be used in point cloud object recognition and classification. First, an L1-central skeleton algorithm with strong robustness is used to extract the skeleton of point cloud data, and a series of skeleton points is obtained. Then, the skeleton points are segmented by using the octree based region growing segmentation method. The selected criteria are normal vectors and residuals. Finally, OpenGL is used to connect the segmented parts. Numerous experiments are conducted with point cloud data of various shapes, such as animal, plant, human, and alphabetic models, and good results are obtained.

This study proposes a detection method for the color structured light stripe based on hierarchical clustering. First, the proposed method converts a color image to a gray image and computes the initial value of central point of the structured light stripe along each row of the image based on hierarchical clustering. Second, the sub-pixel location of central point of the structured light stripe is conducted based on the neighborhood grayscale information, which is indexed on the basis of color distance and Euclidean distance. Finally, the color structured light stripe is detected with image progressive scanning from top to bottom. The proposed method has been validated by both simulations and experiments. The results indicate that the detection precision of the stripe center is 0.47 pixel. The proposed method can still maintain the consistency and integrity of the stripe when the stripe has a tiny fracture. The method can be used in the field of structured light three-dimensional reconstruction and has significant relevance in theoretical research and engineering applications.

A novel structured light technique is proposed in which the strip-edge-based structured light coding strategy is used to generate the projection patterns for reducing the area of oversaturation in an image captured using a camera. The structured light images are captured using both long and short exposure time during a single measurement to avoid image saturation. The long exposure time is used to obtain high-quality images of dark areas, whereas the short exposure time is used to obtain high-quality images of saturated areas. Furthermore, in the short exposure stage, the intensity relation between the projected image and the captured image is estimated by fitting a nonlinear function, and the saturated pixels in the captured image are adaptively adjusted to correspond to the pixel intensities in the projected image. Subsequently, the short- and long-exposure images are combined to form a set of fringe images with a high signal-to-noise ratio, and new images are decoded to achieve accurate three-dimensional measurement. The experimental results denote that the proposed structured light technique exhibits high measurement accuracy for object surfaces denoting a large range of reflectivity variation, including stainless steel surfaces.

In this study, the GH3039 superalloy sheets were treated by using a Procudo200 laser shock peening (LSP) system to study the effect of LSP on the fatigue life of the GH3039 superalloy. Further, the microstructure and residual stress of the samples before and after LSP were characterized by using an optical microscopy and an X-350A residual stress tester, respectively. The fatigue life tests for the samples before and after LSP were performed by using an MTS Landmark 370.10 electro-hydraulic servo test system, and the fracture morphologies of the samples were characterized by using a ∑IGMA500 scanning electron microscope. The experimental results denote that the grain size of the surface layer of the superalloy is refined by LSP. The average residual stress induced by LSP becomes -255.07 MPa at the sample surface, and the surface roughness increases from 0.0346 μm to 0.048 μm. Furthermore, the fatigue life is observed to be improved by 2.41 times after LSP. Finally, the increase in fatigue life of the LSP-treated GH3039 superalloy can be mainly attributed to the residual compressive stress and grain refinement.

The effect of vacuum test condition on the microstructure and properties of the laser cladding Ni-based WC coating is investigated. The results denote that under vacuum test condition, the microstructure of the coating is improved, the coarse dendrites in the coating are reduced, and the structure becomes denser and more uniform. The microcracks in the coating are significantly reduced, and there are almost no porosity defects. The intermetallic compounds in the coating are uniformly dispersed in the Ni-based solid solution, which significantly increases the hardness of the coating. In addition, the friction and wear properties of the coating are significantly improved. The friction coefficient is reduced by approximately 14% when compared with that in the non-vacuum condition, whereas the wear resistance is increased by approximately 56%. Under vacuum test condition, the overall performance of the coating is significantly improved.

To solve the problem of model drift caused by scale variation in visual vehicle tracking, this study proposes a scale search method for the vehicle target based on a kernelized correlation filtering algorithm. The change direction of the target scale is deduced by comparing the average peak related energy of correlation filtering responses obtained from object regions with three given scales, followed by an iterative search for the best scale of the current target in the change direction. To ensure that the correlation filtering template can adapt to the change of vehicle appearance in the process of motion, the template is upgraded with adaptive weight under the condition of the best scale estimation. The manner of adaptive weighting further improves the accuracy of the template. Numerous experiments show that the proposed method effectively solves the problem of model drift caused by scale change in vehicle tracking and provides better tracking performance than other correlation filtering algorithms.

Aiming to improve the tracking accuracy of the correlation filter tracking algorithm when faced with occlusion, background clutter, and deformation of the object target, this study proposes a correlation filter tracking algorithm for multiple features based on temporal consistency and spatial pruning. First, in the training stage, the energy distribution of the filter template is pruned using the binary matrix mask to make the template information more concentrated in the target area, which alleviates the boundary effect caused by the cyclic shifted samples. Second, the l2-norm is used as the temporal consistency model to establish smoothness constraints for the filters of two consecutive frames so that filter templates can learn the context information of consecutive-frame target and increase the anti-interference ability of the algorithm. To further improve the expressive ability of the target template, ResNet50 deep features, which contain rich semantic information, are introduced into the tracking framework. Principal component analysis is used to reduce the dimension of the extracted depth feature, and traditional features in combination with deep features improve the accuracy and robustness of the tracking results. A comparison of the proposed algorithm with five existing algorithms verifies the proposed tracking algorithm’s robustness in dealing with distractors such as target occlusion, background clutter, and deformation.

In this study, we propose a foreground-aware based spatiotemporal correlation filter algorithm based on the spatially regularized discriminative correlation filter (SRDCF) to deal with long-term object tracking failures caused by background clutter, occlusions, and out-of-view objects. Initially, a foreground-aware correlation filtering algorithm is proposed to distinguish the foreground and background of the object accurately. Subsequently, the foreground-aware filter is added to the time regularization term to keep the filter with spatiotemporal regularization function in a low-dimensional discriminative manifold. Simultaneously, the solution based on the alternating direction method of multipliers (ADMM) is conducted to achieve real-time operation of the tracking method in the traditional feature expression. Finally, the activation threshold of object re-detector is determined, and the candidate region method combined with correlation filtering method is used to achieve re-detection, so as to achieve the purpose of long-term tracking. We conduct experiments using traditional and convolutional features with respect to the OTB2013 standard dataset and observe that the average success rates of tracking are 5.6% and 7% higher, respectively, when compared with that of SRDCF. Therefore, the proposed approach is a robust method for handling background blur, rotations, occlusions, and out-of-view objects.

Point cloud registration is a fundamental element of the three-dimensional reconstruction processes. In this study, a point cloud registration algorithm is proposed based on the maximum sum of squares of the correlation coefficients (MCC) to address the issues of scattered point clouds, missing data, and low registration efficiency and accuracy under noise interference. Further, the target point cloud and the point cloud to be registered are de-averaged and rotated, so that the MCC between row vectors of the two sets of point clouds can be achieved after rotation. Subsequently, particle swarm optimization algorithm is used to derive two sets of intermediate-state rotation matrices. Finally, based on these matrix sets, the rotation matrix and translation vector between two point clouds are obtained for registering the point cloud. The simulation results show that the proposed algorithm is faster, more accurate, and more robust compared with the remaining existing algorithms when point clouds are scattered, missing, and interrupted by noise.

Parameter calibration of a line-structured light vision measurement system is the first step required for three-dimensional measurement applications, wherein the most important task is stable acquisition of the feature points on the optical plane. An efficient calibration method is proposed based on a camera projection matrix and a homography matrix considering the characteristics of planar targets as well as the deficiency of the method of obtaining feature points based on the cross-ratio invariant principal. First, the homography matrix is calculated using the corresponding relationship between the feature points on the plane target and their world coordinates. Subsequently, all feature points lying on the intersection of the optical plane and target can be obtained simultaneously based on the camera projection and homography matrices. The experimental results indicate good stability and simple implementation for the new method, as well as high accuracy due to the utilization of all feature points.

In this study, copper oxide (CuO) nanosheet arrays are synthesized on a copper substrate by using the water bath method. The CuO nanosheet arrays formed after a 10-min reaction time exhibit the highest initial discharge specific capacity of 629.1 mA·h·g -1and good cycle properties (79.6% retention after 100 cycles). When compared with the traditional bulk copper oxide anode materials, the CuO nanosheet array structure used as the negative electrode material for lithium ion batteries effectively solves the volume expansion during charging. Furthermore, it reduces the diffusion distance of the lithium ion in the CuO crystal, improving the specific capacity and performance of CuO used as the negative electrode materials.

A liquid crystal spatial light modulator (LCSLM) has a wide range of applications in the non-display field. In this study, the phase modulation characteristics of LCSLM are evaluated using a Shack-Hartmann wavefront detector. The results denote that the linear phase modulation range of the low-order Zernike mode is relatively large, and the diffraction efficiency significantly decreases as the modulation range increases. Furthermore, the possibility of using LCSLM for performing aspherical surface measurements is discussed, which provides a new approach for expanding the method of aspheric surface shape detection.

To study the influence of host material on the performances of organic light emitting diodes, a series of single-emitting layer (EML) devices are prepared by selecting energy-matching electron transport material Tm3PyP26PyB as electron-type host material and comparing it with single-EML devices prepared by using the bipolar host material CzSi. Additionally, TcTa, which has better hole transporting ability, is used as the first EML’s host material while Tm3PyP26PyB and CzSi are used as the second EML’s host materials, respectively, to prepare a series of double-EML devices for comparison. In comparison with Tm3PyP26PyB, CzSi has bipolar properties that make it superior in balancing holes and electrons and in broadening recombination zones. This allows better performance to be obtained from single-EML and double-EML devices. Consequently, the optimized single-EML device using CzSi host material has a maximum brightness, current efficiency, and power efficiency of up to 8634 cd/m 2, 18.70 cd/A, and 16.78 lm/W, respectively; while the maximum brightness, current efficiency, and power efficiency of the optimized double-EML device with TcTa and CzSi host materials are up to 10770 cd/m 2, 30.12 cd/A, and 30.52 lm/W, respectively.

This study focuses on selected soil samples containing different types of lubricating oil, engine oil, diesel oil, and gasoline. Three-dimensional (3D) fluorescence spectra are extracted from different soil samples, and 7 characteristic parameters are calculated for each of them, including the fluorescence intensity mean, standard deviation, transverse and longitudinal coordinates of center of gravity, correlation coefficient, long-axis slope, skewness, and kurtosis. Spectral data are used as identification characteristics for oil. Principal component analysis (PCA) is performed on the 7 characteristic parameters, and the feature vectors of the first 3 principal components after dimension reduction are extracted, accounting for a cumulative contribution rate of 88.79%. Clustering analysis reveals highly similar principal components of 5w-40 and 15w-40 lubricating oils; therefore, these oils can not be accurately classified. Subsequently, the first 3 principal components obtained by PCA are input into the back-propagation artificial neural network and the types of petroleum organic matter are used as outputs for oil identification, resulting in a 95.6% comprehensive recognition rate. Experimental results demonstrate the feasibility of identifying oil pollutants directly using 3D fluorescence spectroscopy of oily soil. Additionally, technical support is provided for subsequent research on oil pollutant identification in soil based on 3D fluorescence spectroscopy, indicating good application prospects.

The building environment in urban areas is complex. Achieving high building detection accuracy from remote sensing images is challenging because of the difficulty associated with distinguishing between buildings and the environmental information. To solve this problem, an improved U-type convolutional neural (U-net) network with enhanced low-dimensional feature information is proposed for detecting buildings from the remote sensing images. Initially, a building is detected using the U-net network model typically employed for medical image segmentation. Further, the low-dimensional information is weakened at each step of the network propagation process. Before merging the feature map of a certain level in the feature pyramid with the feature map of the corresponding expansion path level, it is merged with the feature map of the previous level to optimize the detection accuracy of the building edges. According to the experimental results obtained using a dataset of remote sensing images covering a range of approximately 340 km 2, the proposed method achieves values of 83.9%, 92.8%, and 83.6% for the intersection-over-union (IoU), pixel accuracy, and Kappa coefficient, respectively, demonstrating its superior performance when compared with the fuzzy C-means clustering algorithm, fully convolutional neural network, and classic U-net methods.

A joint localization algorithm based on the threshold comparative weighted (TCW) Taylor series expansion is proposed to address the problem of low positioning accuracy caused by the non-line-of-sight propagation of wireless signals in complex indoor environments. First, the non-line-of-sight error in the measured value of the time of arrival (TOA) is eliminated in real-time by Kalman filter. Then, TCW is used to calculate the location of the target node in the location area determined by the smoothed TOA value and the angle of arrival (AOA) of the signal with measurement noise. The calculated result is taken as the initial value of Taylor series expansion. Finally, the iterative solution is carried out to achieve the second fine positioning. The simulation results demonstrate that compared with the traditional centroid-Taylor series expansion location algorithm and TOA/AOA hybrid location algorithm based on the least-squares method, the initial location results can be more accurate and closer to the Cramer-Rao lower bound by adding AOA constraints and assigning different dynamic weights to different locations.

This study proposes a remote sensing image detection method based on deep learning to solve the issues of human intervention, slow speed, and low accuracy associated with the traditional remote sensing image detection algorithm. A dense connected network is considered to completely use the features extracted from each layer and reduce the network inference time. Further, an expanding block structure with a large perceptive field is adopted, and the low- and high-level feature informations of the network are combined based on the expanding block structure and deconvolution network. Thus, the performance of multiscale object detection for remote sensing images is improved. The experimental results denote that the proposed method exhibits high accuracy and short detection time, especially during the detection of small objects.

A single photon avalanche photodiode detector (SPAD) has many advantages such as large avalanche gain, fast response, high detection efficiency, and easy integration. SPAD array devices can be used for low-light three-dimensional imaging; these devices have important applications in fields such as biochemistry, quantum communication, and lidar. Therefore, it is significant to study the detection technology of SPAD and its array. In this paper, we review and present the working principle and array structure performance of a near-infrared InGaAs/InP SPAD unit. We analyze the major influencing factors such as the dark counting rate, detection efficiency, and after pulses; moreover, we investigate the main direction for device optimization. Further, the main technical schemes of the SPAD array devices used in recent years have been summarized. We provide the sources of crosstalk and methods for eliminating crosstalk. In addition, we compare the technologies used and the results of relevant research institutions.

Herein, the aging and common defects of polyethylene (PE) pipe joints are inspected based on terahertz time-domain spectroscopy (THz-TDS). The reflective indices and absorption coefficients of PE specimens with different aging time are calculated,and the imaging detection for the typical defects of pipe joints is conducted. It is found that the reflective indices of the PE specimens have no obvious change in 0.2-2.0 THz range after aging for a long time. However, the absorption coefficient shows a decreasing tendency with increasing aging time. Finally, the defects are imaged using reflective THz-TDS. Pixels around the defect areas are different from those around non-defect areas in THz images, and the defect position and outline are revealed in the THz images.

Slides are transparent pieces of glass used to hold samples in biology and medicine. To improve real-time detection of their content in production environments, slides are ablated using a 1064-nm laser and the plasma's emission spectra are recorded for different experimental parameters. Herein, the Si I 288.16 nm, Ca I 422.67 nm, and Na I 589.00 nm lines are selected for analysis. The spectral intensity and signal stability are used as indices to optimize the key experimental parameters, i.e., spectral acquisition delay time, laser energy and repetition rate, ablation mode, and lens-to-slide distance. Results show that, for the experimental system of laser-induced slide ablation, the optimal spectral acquisition delay time is 3 μs and the optimal laser repetition rate is 2 Hz. A more stable spectral signal can be obtained using a single-point ablation method. In addition, for a lens with a focal length of 100 mm, the optimal focusing position is 8 mm below the slide surface; hence, the optimal lens-to-slide distance is 92 mm. The energy of each laser pulse is set to 150 mJ to avoid damaging the slide while still obtaining spectra with sufficiently highline intensities and signal stabilities to meet the experimental requirements. Optimizing the experimental system's key parameters in this manner provides an experimental basis for real-time, online, in-situ detection of slide using laser-induced breakdown spectroscopy.

To improve the accuracy of quantitative analysis of laser-induced breakdown spectroscopy (LIBS), the quantitative analysis of Mn in different alloy steels is conducted by LIBS. The calibration model is established by combining support vector regression (SVR) to study the influences of different input vectors in the SVR model, and the internal standard method is used to correct the results. The results show that when the input vectors are the internal standard element correction and signal-to-back ratio, the relative standard deviation and the relative error of the test set are 2.6% and 11.97%, respectively, and the regression effect is optimal. In the quantitative analysis of Mn in the alloy steel, the corrected binary input vector reduces the parameter fluctuation, corrects the matrix effect, and provides a reference for the optimization of the input data for the quantitative analysis of steel elements by LIBS.

A method is proposed to construct metameric pairs for illuminant detection. The proposed method allows the quick and effective determination of the illuminant type. According to this method, a metameric black spectrum set is initially decomposed from the Standard Object Colour Spectra Database under the target illuminant based on the matrix R theory. Subsequently, a metameric sample set of the target sample is obtained by adding a metameric black to any target sample. After eliminating the samples that exhibit negative spectra, the sample with the largest metamerism index is selected to construct a metameric pair, with the general metamerism index as an evaluation index. Further, the proposed method is used to construct metameric pairs under four standard illuminants. The validity of the proposed method can be verified by calculating the color-difference value of the metameric pairs under different illuminants. The experimental results demonstrate that the metameric pairs exhibit a significant color-difference level under non-target illuminants and they can effectively determine the target illuminant type.

Herein, 32 plastic pack belt (rope) samples were examined via X-ray fluorescence spectrometry. At the acquisition time of 80 s, the elemental composition was most stable. The function model was constructed by combining the multivariate statistical method. The ward’s method was selected as the clustering method. The square Euclidean distance was used as the measurement interval to describe the degree of affinity between the samples for hierarchical clustering. The 32 plastic pack belt (rope) samples were divided into four categories. The accuracy and reliability of the hierarchical clustering were tested via correlation between the samples and discriminant analysis, and the internal relationship between the contents of the sample elements was determined. In this experiment, 32 samples of plastic pack belts (ropes) were classified using the hierarchical clustering method, which promoted the application of pattern recognition technology for physical and chemical examination.