This paper presents an improved algorithm to solve the problem of ghosting in the visual background extractor (ViBe) algorithm for pedestrian detection. The initialization strategy of the ViBe algorithm is improved using the YOLO v3-spatial pyramid pooling (SPP) algorithm to eliminate ghosts. Thus, the YOLO v3-SPP algorithm detects pedestrians in the first frame, eliminates the detected pedestrians, and replaces the first frame of the ViBe algorithm with the output image to eliminate ghosts. The results of the analysis and verification show that the algorithm can effectively solve the ghost problem.

Aiming at the problems that infrared and visible images are prone to artifacts and unclear outlines of small targets after fusion, an infrared and visible images fusion algorithm based on the combination of multi-scale features and attention model is proposed. The feature maps of different scales of the source image are extracted through five times of down-sampling, and then the infrared and visible image feature maps of the same scale are input to the fusion layer based on the attention model to obtain an enhanced fusion feature map. Finally, the small-scale fusion feature map is up-sampled five times, and then added to the feature map of the same scale after up-sampling, until the scale is consistent with the source image, and the multi-scale fusion of the feature map is realized. Experiments compare the entropy, standard deviation, mutual information, edge retention, wavelet feature mutual information, visual information fidelity, and fusion efficiency of fused images under different fusion frameworks. The method in this paper is superior to the comparison algorithm in most indicators, and the target details are obvious and the outline are clear in the fused images.

In view of the current target tracking algorithm, it is difficult to effectively track the target when it is blurred or occluded; therefore, an anti-occlusion algorithm based on multi feature adaptive fusion is proposed in this study. First, the gradient direction histogram feature HOG and color histogram feature are extracted, and the correlation filtering response of the two features is adaptively fused to maximize the tracking quality. In the tracking process, a high-quality filter template is stored according to the quality of the response map, and the quality difference between the high-quality template and normal update template is used to detect the occlusion of the target. When the target occlusion disappears, the template of the tracker traces back to the high-quality template to retrack the target. According to the experimental results for OTB100 and UAV123, this algorithm has a better performance than other similar correlation filtering algorithms and can still track well in the case of target occlusion.

Aiming at the problems of low contrast and blurred details of images captured under low-light conditions due to the influence of light and the environment, important information is lost, and an image enhancement method based on grayscale transformation and improved Retinex is proposed. First, the global grayscale transformation function optimized by the gravity search algorithm(GSA) is used to perform grayscale transformation on the grayscale image of each RGB channel of the image to enhance the image illumination intensity and make it closer to the uniform illumination scene. The image is then converted to the HSV color space. The V channel (luminance channel) is processed by the improved multi-scale Retinex (MSR) algorithm; range-based adaptive bilateral filtering and Gabor filtering are used as the surround function of the Retinex algorithm, and the characteristics of the two filters are combined to enhance the brightness and detail of the image. Finally, gamma correction is used to avoid image color casts caused by image fusion. The experimental results show that the enhanced image processed by this method is better than that processed by other methods in subjective and objective evaluation, the color distortion of the image is smaller, and the details are clearer, which paves the way for the subsequent application of the image.

Due to the lack of image saliency preserving in the existing fusion rules, a self-attention-guided infrared and visible light image fusion method is proposed. First, the feature map and self-attention map of the source images are learnt by the self-attention learning mechanism in the feature learning layer. Next, the self-attention map which can capture the long-distance dependent characteristics of the image is used to design average weighted fusion strategy. Finally, the fused feature maps are reconstructed to obtain the fused image, and the learning of image feature coding, self-attention mechanism, fusion rule, and fused feature decoding are realized by generative adversarial network. Experiments on TNO real-world data show that the learned self-attention unit can represent the salient region and benefit the fusion rule design, the proposed algorithm is better than SOAT infrared and visible image fusion algorithms in objective and subjective evaluation, and it retains the detailed information of visible images and infrared target information of infrared images.

To improve the detection accuracy of methane concentration, an experimental system based on tunable diode laser absorption spectroscopy(TDLAS) technology was built. Taking advantage of the high absorption intensity of methane at a wavelength of 1653.72 nm and its ability to eliminate the interference of other gases to the greatest extent, methane concentration was detected by extracting the second harmonic signal. The heursure hard threshold algorithm, heursure soft threshold algorithm, and sqtwolog fixed threshold algorithm are used as the wavelet transform threshold algorithms, respectively; the sqtwolog fixed threshold algorithm is preferred as the wavelet transform threshold algorithm by analyzing the methane absorption signal spectrum, the methane second harmonic signal spectrum, the signal-to-noise ratio and root mean square error of the methane absorption signal obtained without denoising and after denoising. The results of the linear fitting experiment of methane standard gas with different concentrations and the repeatability experiment of methane standard gas of a specific concentration show that the noise interference can be effectively reduced by the wavelet transform using the sqtwolog fixed threshold algorithm. The goodness of fit R2 between the second-harmonic signal extracted after denoising and the real methane concentration was 0.984, indicating that the fitting effect was better. TDLAS technology combined with the wavelet transform denoising algorithm can realize the detection of methane concentration and improve the detection accuracy of methane concentration. TDLAS technology combined with the wavelet transform denoising algorithm can realize the detection of methane concentration and improve its detection accuracy.

PoTo address the limitations of current dry eye diagnostic equipment in the market, such as high cost, large volume, and low accuracy of detection results due to the use of the frame difference method, this study proposes portable dry eye diagnostic equipment based on near-infrared images. The device, which uses a Windows computer based on near-infrared imaging, is portable and can be used to detect dry eye symptoms such as the tear film break-up time (TBUT), meibomian gland (MG), and tear meniscus height (TMH). The fractal box dimension method was used to detect TBUT to avoid the inaccuracy of the frame difference method. MG was detected using near-infrared (850 nm) fill light imaging technology, and the contrast limited adaptive histogram equalization (CLAHE) algorithm was used to highlight the gland region, which can detect the area missing rate more accurately. To verify the accuracy of the test results, 50 samples were tested and compared with the CSO Antares instrument from Italy and the ICP OSA from SBM. The correlation coefficient between the TBUT and TMH measurements of the device and the results of the two controls was P<0.05, whereas the detection results were consistent; the accuracy and specificity of MG were 86% and 84%, respectively. The experimental results showed that the device can be used as a screening and diagnostic device for dry eye disease in eye and vision optics centers.

A method for calculating the flare area based on image segmentation is proposed to study the ultraviolet image characteristics of the wet-contaminated discharge of disk-shaped insulators. First, based on the C-V model, the radial distribution of an ultraviolet image of a disk-shaped insulator is studied. It is then compared with the infrared temperature-field curve to verify its rationality. Finally, the ultraviolet image characteristics of the wet-contaminated insulators with different electrical resistivities and wind speeds are studied. The results show that the C-V model has strong anti-interference performance when applied to image processing, and is suitable for the study of the radial discharge intensity of disk-shaped insulators. The ultraviolet flare area of the insulator increases with a decrease in the surface resistivity. The wind speed has an obvious influence on the discharge intensity of the disk-shaped insulator. The results provide a reference for evaluating the radial discharge intensity of disk-shaped insulators.

The recent progress in the LPE growth of HgCdTe(MCT) epi-layer at the Kunming Institute of Physics (KIP) is reported. The precipitates and inclusion density of the CdZnTe substrate are less than 5×103 cm.2 and the etch pit density (EPD) is less than 4.0×104 cm.2. Owing to the oriented growth technology of the CdZnTe single crystal with..120 mm, the variation of the Zn components in the prepared substrate wafers with..120 mm (111) was less than 0.36%. The maximum size of the HgCdTe films grown by LPE was up to 70 mm×75 mm, with an etch pit density (EPD) of 5×104 cm.2 and DCRC-FWHM less than 35 arcsec (could be lower than 25 arcsec). Over a substrate area of 50 mm×60 mm, the thickness homogeneity of the LPE MCT epi-layer is better than.d=.1.25.m, and a cut-off wavelength homogeneity better than..=.0.1.m is achieved for the LWIR MCT epi-layers; that of the MWIR epi-layers are.1.m and.0.05.m, respectively. Due to the improvement of material technology, which effectively increases the yield of low-cost MCT IRFPAs at KIP, high-performance LW/VLW/HOT IRFPAs and large array detectors such as 2048×2048 and 4096×4096 based on MCT/CZT have been developed.

Considering the demand for high-precision detection of SWIR remote sensing systems, the demand for the quantitative application of SWIR detection has increased in recent years. The change in inhomogeneity caused by shortwave infrared remote sensor was analyzed. The design scheme of non-point source blackbody calibration combined with star calibration in-orbit absolute radiation calibration; the actual development process of a remote sensing task; and the main influencing factors and optimization measures, including the optimization of the satellite calibration scheme, optimization of satellite blackbody temperature control, and star extraction algorithm, were analyzed. Based on the results of laboratory tests and estimation of the accuracy of the absolute radiation calibration, the evaluation results show that the calibration accuracy can meet the application requirements.

To address the problems of poor infrared image segmentation and slow speed in the online monitoring system of power equipment, an improved infrared image segmentation algorithm based on the Chan-Vese model is proposed. First, by introducing the edge energy term, the local control ability of the model is enhanced and the contour shift is effectively suppressed. Second, a radial basis function is used to replace the traditional length regularization term, which simplifies the calculation. Subsequently, the initialization process is omitted by introducing internal energy items, which reduces the running time of the algorithm. After the experimental verification, the average DSC was 0.9808, the average value was 0.025, and the algorithm running time was 66.8% lower than the overall average of the other models. The improved Chan-Vese model segmentation algorithms DSC and RSE are better than the GAC-CV, CV-RSF, regional level set, and multiphase-CV model segmentation algorithms.

To study the effect of the luminance gain of a third-generation low-light-level (LLL) image intensifier on the image quality, we compare and analyze the image quality of the output image under different gain conditions to improve the image quality of a third-generation image intensifier. First, based on the theory of the third-generation LLL image intensifier, it is proven that the luminance gain directly affects the image quality of the image intensifier. Then, via signal-to-noise ratio (SNR) and resolution, two important parameters of the image quality evaluation, image quality evaluation system, and experimental equipment, are set up. Finally, the experimental results show that under moonless illumination, when the luminance gain is at the optimal value, the resolution of the output image is increased from 32 lp/mm to 40.3 lp/mm while the field of view is bright and clear. It has been proven that this research has guiding significance in how to obtain the best image quality of LLL night vision instruments by reasonably setting the luminance gain value in the night environment.

With the improvement and promotion of the Mercury Cadmium Telluride (MCT) material preparation process, the dark current of the FPA is suppressed to a certain extent, and an increase in the operating temperature of IR detectors has become a development trend. The development of high-operating-temperature IR detectors promotes the development of micro-Stirling cryocoolers to reduce the size, weight, power consumption, and cost, and promote their performance. The SWaP3 design concept of Stirling cryocoolers for HOT IR detectors is presented. Design techniques such as thin-walled tube short cold fingers, high-efficiency small-size controllers, comprehensive thermal management, reliability prediction, and the recent domestic and foreign development status of Stirling cryocoolers for HOT IR detectors are summarized.

Infrared fault images have the limitations of a low recognition accuracy and low detection rate in a PV module inspection task using a UAV. To address these issues, a feature enhanced YOLO v5s fault de-tection algorithm is proposed. First, the loss function is optimized, the original regression loss calculation method is changed from GIOU to EIOU, and the confidence loss balance coefficient is adjusted adaptively to improve the model training. The InRe feature enhancement module is then added before each detection layer to enhance the ability of the target feature extraction by enriching the feature expression. Finally, compara-tive experiments are conducted using the infrared photovoltaic dataset created in this study. The experi-mental results show that the detection mAP of our method is 92.76%, whereas the detection speed is 42.37 FPS. The mean average precisions of the hot spot and component falling off were 94.85% and 90.67%, re-spectively, which can fully meet the requirements of the automatic inspection of the UAV.