As a new photoelectric material, black silicon has been widely studied in photovoltaic solar cells, photodetectors, CMOS image sensors and other fields. Among them, the photoelectric detection technology of black silicon has attracted much attention, and important research progress has been made in recent years. In this review, the structure of black silicon materials has been firstly introduced, then the properties of black silicon materials prepared by femtosecond laser etching, wet etching and reactive ion etching are briefly discussed. Secondly, the structure and performance of different black silicon photodetectors based on the above preparation methods are summarized, then the application of black silicon devices in different fields is discussed. Finally, the photoelectric detection technology of black silicon is analyzed and prospected, and the future development direction of black silicon materials and devices is discussed.

Infrared ZnS bulk material is widely used in domes, infrared lens and windows. The fabrication technology of ZnS bulk material is reviewed including hot press (HP) and chemical vapor deposition + hot isostatic press (CVD+HIP). The influence of fabrication process on optical properties is analyzed. It is concluded with the technology trends prospects for the future development of bulk ZnS bulk material.

In terms of the problems of low brightness, high noise and blurred edges in low-illumination images, based on Xilinx's Artix-7 series FPGA chip, the XQE-1310 image sensor with good low-light performance is driven to filter and edge the video signal collected by the detector. After the detection, a series of operations such as acquisition and processing of low-light images were completed, and the processed video signals were displayed in real time through the CameraLink video format. Finally, a low-light night vision system was designed. The experimental results show that the minimum working illuminance of the system can reach the order of 10-2 lx. The filtering algorithm can effectively filter the salt and pepper noise in the image while maintaining the edge information of the image. The adaptive edge detection algorithm can adjust the threshold in real time according to the illuminance level. It highlights the contour information of objects in low-light environments. The system makes full use of the advantages of fast speed and high efficiency of FPGA (Field Programmable Gate Array), and the final imaging result is clear and stable, which is convenient for human eyes to observe.

The ultraviolet (UV) image intensifier is the core device of an UV imaging system. Its imaging quality determines its ability to detect and image UV optical signals. The modulation transfer function (MTF) represents the system's ability to transfer information at different frequencies and is an objective indicator of image quality assessment. Based on the MTF test principle of slit imaging and the Fourier analysis method, a set of MTF test systems for UV image intensifiers is established in this study. MTF test experiments were performed on three UV image intensifiers. The MTF curve cutoff frequencies of the three UV image intensifiers were between 32 and 34 lp/mm, and the imaging quality of the three UV image intensifiers was compared based on the MTF curves. Finally, the standard deviations of the MTF values of several important frequency points obtained from repeated tests were lower than 0.02.

To effectively solve the problem of low detection rates of dim and small targets caused by dynamic background changes, a detection method based on spatio-temporal filtering is proposed in this paper. Based on an analysis of the imaging characteristics of infrared images, an improved anisotropic spatial filtering algorithm is proposed to evaluate the difference in various gradient characteristics of the target area, background area, and edge contour area. The algorithm fully utilizes the gradient information in the spatial domain to construct the diffusion filter function in different directions. According to the gradient difference in various characteristics of the image, the mean value of the two directions with the smallest value of the diffusion function is selected as the result of spatial filtering to retain the target signal to the maximum extent. To effectively enhance the energy of dim and small targets and address the shortcomings of high-order cumulants that only use the temporal domain information of pixel points for energy enhancement, an energy enhancement algorithm based on spatial-temporal neighborhood blocks is proposed. Experimental results reveal that the proposed algorithm can effectively enhance the detection of dim and small targets in dynamically changing scenes.

Polarized light reflection information can directly invert the intrinsic characteristics of a target and has strong anti-interference characteristics in the transmission process. Thus, polarization imaging technology can be applied to the fields of intelligent monitoring and traffic monitoring in various complex environments. In recent years, deep-neural-network methods for interpreting image detection targets have been developed rapidly and widely used in various fields of image processing. In this study, a vehicle multi-target detection algorithm based on polarized images and deep learning is proposed. First, the target polarization degree image can be obtained by acquiring the polarization image in real time and analyzing the polarization information. Second, to enhance the high contrast between the detection targets and the background in the polarization image, channel attention and spatial attention are introduced into the backbone network to improve the ability of the network features to perform adaptive learning. In addition, the K-means algorithm is used to perform clustering analysis on the target location information, thereby increasing the network's learning speed in the polarization image and improving the progress of target detection. The experimental results show that this method is effective and fast for vehicle detection in complex scenes with low illumination. This method combines the advantages of polarization imaging and deep-learning target detection and has substantial application scope in road vehicle target detection, recognition, and tracking.

To address the problems of local blur and incomplete background information in the traditional fusion algorithm of infrared and visible images, a new fusion algorithm is proposed in this paper. The edge detection operator was used to extract the image contour, and weighted fusion based on energy was also executed. The similarity between regions was used to extract the signal domain. Finally, image fusion is performed according to the over-signal strength. To verify the correctness of the algorithm, a comparative test was conducted and a quantitative analysis was performed using three parameters: standard deviation, information entropy, and average gradient. Compared with the traditional weighted average algorithm, the standard deviation of this method was up to 106.3 %. The test results confirmed that the fusion method proposed in this study has a better fusion effect and practical value.

To increase the information of the fused image, this paper proposes an improved multi-modality image fusion algorithm that combines the complementary advantages of the non-subsampled shearlet transform (NSST) and discrete wavelet transform (DWT). NSST was used to decompose the two source images in multiscale and multi-direction to obtain the corresponding high-frequency and low-frequency sub-bands. The low-frequency sub-bands were further decomposed into low-frequency energy sub-bands and low-frequency detail sub-bands by the DWT, and the low-frequency energy sub-bands were fused by the maximum selection rules. An adaptive pulse-coupled neural network with improved connection strength (ICSAPCNN) was used to fuse the detailed sub-bands and high-frequency sub-bands, and the energy sub-bands and detailed sub-bands were fused by inverse DWT to obtain the fused low-frequency sub-bands. The NSST inverse transform was used to reconstruct the fusion image with rich details. The experimental results verified that the proposed algorithm is superior to the other algorithms in both subjective vision and objective evaluation and can be applied to the fusion of both infrared and visible source images and medical source images.

Using the traditional algorithm to meet the detection requirements of interference factors, such as complex background and noise, relying on the precise separation and information extraction of infrared targets and environmental background, is difficult. This paper presents a dim–small target detection method for an infrared imaging algorithm based on the improved YOLOv5 for complex backgrounds. Based on YOLOv5, an attention mechanism is introduced in the algorithm to improve the feature extraction ability and detection efficiency. In addition, the loss function and prediction box screening method of the original YOLOv5 target detection network are used to improve the accuracy of the algorithm for infrared dim–small target detection. In the experiment, seven sets of infrared dim–small target image datasets with different complex backgrounds are selected, the data are labeled and trained, and an infrared dim–small target detection model is established. Finally, the accuracy of the algorithm and model is evaluated in terms of the model training and target detection results. The experimental results show that the model trained by employing the improved YOLOv5 algorithm in this study has a significant improvement in detection accuracy and speed compared with several target detection algorithms used in the experiment, and the average accuracy can reach more than 99.6%. The model can effectively detect infrared dim–small targets in different complex backgrounds, and the leakage and false alarm rates are low.

The time-of-flight (ToF) 3D imaging method suffers from reduced precision in the depth measurement of target objects because of multipath interference and mixed pixels. Traditional methods improve the accuracy of the measurement by optimizing and reconstructing abnormal point cloud data or filtering noisy point cloud data. However, these methods are complex and can easily lead to excessive smoothing. The relationship between a valid point cloud and noisy point cloud in a 3D point cloud image is difficult to express using a mathematical model. To address these problems, a point cloud denoising method based on the confidence level is proposed in this paper. First, the probability correlation of multi-frame point cloud data is analyzed, and the confidence level of the point cloud data is used as the basis to distinguish valid point clouds from noisy point clouds. Second, by utilizing the vector duality between multi-frame point clouds, a fast algorithm for extracting point clouds with different confidence levels is presented, and its time complexity is O(L). Finally, the algorithm is used to extract the point cloud data with a high confidence level in multi-frame 3D images to obtain the real measurement data of the target object. We focus on the comparative experiments of four groups of point cloud data in different scenes. The experimental results show that the algorithm can not only effectively filter the noise but also significantly improve the distance measurement accuracy of the target object and enhance the characteristics of the target object; therefore, it has extensive application value.

Both infrared and visible images have certain limitations, and relying on individual types of images cannot meet the practical needs of engineering. Instead, high-quality fused images can be obtained by introducing image fusion techniques. To better guarantee the diversity of the output information features, this study introduces a dual-branch convolutional neural network to achieve local fusion of infrared and visible images. Based on the dual-branch convolutional neural network, red and blue features are obtained from infrared images and visible light images simultaneously, thereby increasing the amount of information in the fusion image. The integer wavelet transform method is used for image compression. When the color-space model is built, the value of the t-factor is adjusted to obtain an ideal fusion image. The experimental results show that the edge information of the image after the fusion of this method is fully preserved, image detail information is enhanced, and fusion effect of infrared and visible images is improved, compared with the existing methods.

To enhance the anti-jamming performance of ship detection, an effective and stable single-stage ship detection network is proposed in this study. The network is composed of three modules: feature optimization, feature pyramid fusion, and context enhancement modules. The feature optimization module extracts multi-scale context information and further refines the high-level feature input characteristics, to enhance the performance of dim–small object detection. The feature pyramid fusion module can generate semantic information with stronger representation ability. The context enhancement module integrates local and global features to enhance the network feature expression ability, reduce the impact of a complex background on detectability, adjust the imbalance between the foreground and background, and eliminate the impact of scale-wave. To verify the effectiveness and robustness of the proposed method, qualitative and quantitative verifications are performed on a self-built dataset. Experimental results show that the proposed network achieves optimal performance compared with the latest benchmark comparison model and considerably improves the detection accuracy without increasing complexity.

Solar photovoltaic power generation is an important component of a country's energy structural adjustment. With the rapid expansion of the scale of the photovoltaic power generation industry in recent years, the need for an automated daily maintenance of photovoltaic power stations has increased. Traditional manual detection methods are inefficient because photovoltaic power stations are spread over a large area. In this study, we investigate the intelligent inspection technology of a photovoltaic power station based on an unmanned aerial vehicle (UAV). A technical route for an intelligent inspection of a UAV-based photovoltaic power station is proposed. We achieve the automation of photovoltaic panel image data acquisition and analysis and investigate defect detection based on computer vision. We realize photovoltaic panel defect detection based on infrared images using an adaptive dynamic threshold method combined with image enhancement technology, facilitating the classification of defects to be determined by using visible light images. The defect locations are further calculated by combining the POS data and the camera model. Finally, we verify the effectiveness of the proposed technical route in an actual scenario.