Although infrared polarization imaging systems have been developed rapidly and widely, a model for evaluating their performance has not been sufficiently developed. Performance models that can match advanced polarization imaging systems are urgently required. Regarding the similarity between the training process of a deep learning network and the process of extracting cognitive information from the human brain, this paper introduces a deep learning method in the field of system performance modeling for the first time and proposes a performance model for infrared polarization imaging systems that can automatically evaluate system performance based on two-dimensional images. The model includes two main modules: a degradation module and a performance awareness module. When evaluating a new system, high-quality original images are input and sequentially passed through an imaging system degradation module, customized according to the hardware parameters of the system, and input into a performance awareness module to obtain the final target acquisition performance. Moreover, to verify the effectiveness of the model, we realized a self-built infrared polarization dataset for sea surface scenes based on infrared radiation theory, and trained and tested the networks. The results obtained when the model was applied to evaluate the performance of infrared polarization imaging systems showed good agreement with subjective perception.

Fuzzy theory is considered to address the uncertainty of infrared image segmentation of electrical equipment, and a new algorithm based on fuzzy inference for infrared image segmentation of electrical equipment is proposed in this paper. First, the intensity, global fault probability, and local grayscale features were defined using the pixel grayscale of the fault region in the infrared image of the electrical equipment, Mahalanobis distance between pixel points, image center of mass, and image dilation operation. Subsequently, 27 fuzzy rules were formulated based on the fuzzy language values of the features, and an infrared image segmentation algorithm based on fuzzy inference was designed. Finally, the algorithm was compared with the traditional Otsu and FCM algorithms in terms of subjective and objective evaluation indexes. Further, the experimental results show that the segmentation accuracy and misclassification error of the proposed algorithm are better than those of the other two algorithms. The algorithm can filter out interference regions with high luminance in infrared images, and exhibits a better segmentation effect on infrared images with small luminance differences and small fault areas.

A method of positioning and integral segmentation of faulty equipment in infrared images acquired during the process of infrared monitoring of electrical equipment in substations with complex backgrounds is proposed to contribute to solving problems including inaccurate positioning and difficult segmentation of faulty equipment. First, the image was segmented using the SLIC superpixel algorithm and the superpixel block was transformed into the Lab color space. The faulty area was obtained after the fault was determined based on the threshold value. Second, relatively bright spots with the maximum connectivity in the image, including faulty equipment, were selected as the original seeds. The number of seeds was controlled based on the principle of maximum variance between classes. Accordingly, primary equipment was obtained using an improved regional growth method. Finally, the overall segmentation of the faulty electrical equipment was completed through an intersection calculation between the faulty area and the main equipment. The results show that the positioning and overall segmentation of faulty electrical equipment under complex backgrounds can be successfully completed using the proposed method. Compared with other segmentation methods, identification of faulty electrical equipment using this method is more complete and accurate.

In order to reduce the influence of changes such as flotation bubble merging and breaking on the foam surface flow feature extraction, a foam surface flow rate detection method based on infrared target segmentation and improved SURF matching in NSST domain is proposed. First, two adjacent froth infrared images are decomposed through NSST, and boundary, brightness, and saliency constraint terms of the graph cut energy function are constructed in the multi-scale domain to realize the segmentation of the merged and broken bubbles. Then, SURF feature point detection are performed on the segmented background region. The main direction of the feature point is determined by statistical the scale correlation coefficients in the sector area, and the multi-directional high-frequency coefficients in the neighborhood of the feature point are used to construct the feature descriptors. Finally, feature points are matched for two adjacent froth infrared images, and the magnitude, direction, acceleration and disorder of foam flow velocity are calculated based on the matching results. The experimental results show that the method in this paper can effectively segment the merged and broken bubbles with high segmentation accuracy, improve the matching accuracy of SURF algorithm, reduce the impact of the bubbles merging and breaking on the flow velocity detection. Compared with the existing methods, the method in this paper improve the detection accuracy and efficiency, which can accurately characterize the flow characteristics of the foam surface under different working conditions and lay the foundation for the subsequent working condition identification.

To solve the problems of large parameters, high complexity, and poor detection performance of multiscale targets in the existing infrared target detection algorithms based on deep learning, a lightweight infrared target detection algorithm for multiscale targets is proposed. Based on YOLOv3, the algorithm uses the MobileNet V2 backbone network, simplified spatial pyramid structure (simSPP), anchor-free mechanism, decoupling head, and simplified positive and negative sample allocation strategies (SimOTA) to optimize the backbone, neck, and head, respectively. Finally, LMD-YOLOv3 with the model size of 6.25 M and floating-point computation of 2.14 GFLOPs was obtained. Based on the MTS-UAV data set, the mAP reached 90.5%, and on the RTX2080Ti dataset, the FPS reached 99. Compared with YOLOv3, mAP increased by 11.7%, and the model size was only 1/10 of YOLOv3.

Because the infrared focal plane detector is limited by manufacturing technology, the image is inevitably nonuniform. The traditional neural network algorithm solves the "ghost" problem using the guided filtering image as the expected template to prevent image edge smoothing by the filter. When the scene is moving, the nonuniformity correction parameters are continuously updated using the time-domain iteration strategy. To suppress the common ghosting phenomenon in the algorithm, an adaptive learning rate was designed based on a combination of the spatial local variance and the time-domain scene change rate, and the threshold was adjusted adaptively using the correction parameters before and after. Simulation results show that the root mean square error of the proposed algorithm is reduced by 45.45% compared with that of the traditional algorithm, and the proposed algorithm can suppress the "ghost" phenomenon well while correcting image nonuniformity.

This study proposed a multisource fusion network (MF-Net) that combines visible and infrared images for the inspection of a photovoltaic panel to achieve photovoltaic panel defect detection, defect classification, and localization. The limitations of the traditional methods include low efficiency, low accuracy, and high cost. In this study, a defect detection network was designed based on the backbone of YOLOv3-tiny. Deep layers are added to the network, constituting a dense block structure to augment semantic information on fused feature maps. The detection scale of the network was extended to improve its applicability at different scales. In addition, an adaptive weight fusion strategy was proposed to achieve feature map fusion, where the fusion coefficients can be allocated according to the pixel neighborhood information. Compared with the backbone, the results show that the mAP of our network improved by 7.41%. The performance improves (by approximately 2.14% mAP) when the weighted fusion strategy is replaced with ours, and the significance of the features can be effectively improved. Relative to other networks, the proposed network that takes the fused images as the input has the highest performance in terms of the F1 score (F1=0.86).

Owing to technical limitations regarding the device and process, the resolution of infrared images is relatively low compared to that of visible images, and deficiencies occur such as blurred textural features. In this study, we proposed a super-resolution reconstruction method based on a deep convolutional neural network (CNN) for infrared images. The method improves the residual module, reduces the influence of the activation function on the information flow while deepening the network, and makes full use of the original information of low-resolution infrared images. Combined with an efficient channel attention mechanism and channel-space attention module, the reconstruction process selectively captures more feature information and facilitates a more accurate reconstruction of the high-frequency details of infrared images. The experimental results show that the peak signal-to-noise ratio (PSNR) of the infrared images reconstructed using this method outperforms those of the traditional Bicubic interpolation method, as well as the CNN-based SRResNet, EDSR, and RCAN models. When the scale factor is ×2 and ×4, the average PSNR values of the reconstructed images improved by 4.57 and 3.37 dB, respectively, compared with the traditional Bicubic interpolation method.

To address the low detection accuracy and poor robustness of infrared images compared with visible images, a multiscale object detection network YOLO-MIR(YOLO for multiscale IR images) for infrared images is proposed. First, to increase the adaptability of the network to infrared images, the feature extraction and fusion modules were improved to retain more details in the infrared images. Second, the detection ability of multiscale objects is enhanced, the scale of the fusion network is increased, and the fusion of infrared image features is facilitated. Finally, a data augmentation algorithm for infrared images was designed to increase the network robustness. Ablation experiments were conducted to evaluate the impact of different methods on the network performance, and the results show that the network performance was significantly improved using the infrared dataset. Compared with the prevalent algorithm YOLOv7, the average detection accuracy of this algorithm was improved by 3%, the adaptive ability to infrared images was improved, and the accurate detection of targets at various scales was realized.

Aiming at the problem of infrared radiation in the 8-14 μm band of ships, a water curtain spray method was adopted to attenuate the infrared radiation intensity of a target, and an infrared thermal imager was employed to build a wall attached nozzle test system. Through comparative design experiments, the influences of different spraying distances, total spray flow rates, and temperatures of the steel plates on the infrared cooling characteristics of the wall nozzles were analyzed. The test results show that when the target steel plate is in the hollow area, and the boundary between the hollow area and the coverage area, the infrared cooling rates are 2.03 and 3.31℃/min, respectively. When the target steel plate was in the coverage area, there were non-overlapping and overlapping areas in the spray. The maximum cooling rate of the non-overlapping area was 6.18℃/min, and the cooling rate of the overlapping area under the same radius was 6.54℃/min. The infrared cooling time in the overlapping zone was 40 s, which was 32 s shorter than that in the non-overlapping zone. Moreover, an increase in the total spray flow resulted in a significant increase in cooling of the steel plate. The higher the initial temperature of the steel plate, the higher the cooling rate of the wall nozzle. In addition, the temperature difference between the steel plates inside and outside the water curtain can reach up to 8.49℃. Studies have shown that the water curtain spray formed by a wall-attached nozzle can effectively cover the surface temperature of the steel plate, reduce the infrared detectability of the ship surface, and realize water curtain spray stealth. These results provide a technical reference for improving the infrared stealth performance of ships.

To solve the difficulty of selecting an appropriate heat sink for complex heat flows, the design of high-power heat sources for space cameras was investigated based on the design principle of reducing the total heat flow to a heat sink. First, the heat flow to a camera was analyzed according to a space environment. Subsequently, by analyzing the heat flow and working mode of the heat source, the efficiency of heat dissipation from the heat source and the area of the radiant cooling plates were reduced by installing radiant cooling plates on both sides of the camera and coupling them through heat pipes. Finally, the thermal analysis was verified using a thermal simulation software based on the camera's space environment and the thermal control measures taken. The simulation results showed that the temperature of the visible focal plane component was -1.9℃ to 12.9℃, the temperature of the infrared camera circuit board was -1.7℃ to 6.7℃, the temperature of the hot end of the chiller was -12℃ to 0.3℃, and the temperature of the chiller compressor was -11.3℃ to 21.3℃. The temperature index requirements were satisfied, and the problem of heat dissipation from the high-power heat source of the camera under complex heat flow was solved.

To realize the low-cost and athermalization design of medium-wave infrared seekers, a high-resolution medium-wave infrared imaging guidance optical system with infrared imaging guidance was designed with low-cost and wide-temperature-range athermalization. The general layout is a two-axis frame; the system chooses one imaging configuration with three pieces of lenses based on Si/Ge material. The detector chooses a Stirling-cooled 640 pixel×512 pixel detector with the pixel size of 15 μm. The prototype design results show that the optical system focal length is 100 mm, field size is 10°×8° at 33 lp/mm, the axis view of the modulation transfer function (MTF) is not less than 0.6, 0.7 field of the off-axis modulation transfer function (MTF) is not less than 0.40, the system distortion is less than 1%, and the efficiency of the cold stop is 100%. Moreover, the narcissus of the system is almost elimination based on pertinence optimization design with fairing. In the temperature range of -40 to 70℃, good image effect was realized. The optical system has the advantages of a simple structure, ease of processing and adjustment, and high yield rate; the imaging quality of the optical system is excellent, and the performance indexes meet the technical specifications.

With the progress of science and technology, cutting-edge products and advanced photoelectric systems have increasingly higher requirements regarding the imaging quality of the optical system. Optical aspheric elements, which are widely used, can effectively correct the aberration, reduce the number of optical elements required by the system, and reduce the weight of the system. Because of the specific surface characteristics, machining and testing such systems are more difficult than for spherical particles; the testing accuracy directly determines the processing accuracy, and the importance of aspheric testing technology is obvious. Herein, the testing technology of optical aspheric surface is summarized according to measuring principle; As direct surface profilometry is widely used in optical aspheric surface machining, combined with the latest testing methods, the measurement technology of aspheric surface direct surface profilometry is mainly introduced; The application of freeform surface and surface profilometry in freeform surface testing, which has attracted increasing attention in recent years, is introduced; Finally, the present situation and development trend of aspheric surface testing technology are summarized.

Space science instruments use solar-guide mirror pointing and tracking systems based on four-quadrant photodetectors to achieve precise pointing control of the sun. To satisfy the requirements of high precision and stability, a four-quadrant photodetector screening method for space applications was proposed, and a screening system for the four-quadrant photodetector was developed. By comparing the dark current, responsivity, and quadrant responsivity uniformity of the four-quadrant detector before and after the screening test, the space-environment adaptability of the detector was analyzed according to the discriminant criteria. Additionally, detectors with early failures or large changes in performance can be eliminated. The results show that the developed screening system has high accuracy, and the equivalent input-current noise at the front end of the system is 0.58 fArms. After the screening test, the maximum absolute value of the dark current of each channel of the detector selected according to the evaluation standard was 6.08 pA. The maximum change in the response of each channel was 0.716%, and the maximum change in the response of each quadrant before and after the nonuniformity screening was 1.24%. Finally, the four-quadrant detector was applied to the solar guide mirror pointing and tracking system, and met the requirements of aerospace environmental conditions. The screening device and screening method can be employed to screen four-quadrant photodetectors for space applications, providing significance reference for the screening of other photoelectric devices.

An EMCCD camera was introduced to realize 25 fps continuous dynamic imaging with 640×512 resolution under 1×10^-3 lx illuminance. Through the construction of the camera hardware platform, as well as the analysis of the EMCCD working timing, AFE working timing, BT.656 encoding, and Camera Link encoding timing, the camera uses FPGA and HDL to generate the corresponding driver timing. This includes EMCCD exposure and readout, AFE-correlated double sampling and optical dark clamping, analog video progressive to interlacing and stretching, and Camera Link parallel to serial conversion. The camera operates under the following conditions: 1×10^-3 lx simulated night sky illuminance, 1000× EMCCD gain, 25 mm lens focus, and F1.4; the experimental results demonstrate the imaging frame rate of 25 fps and SNR of 21.8 dB.

For infrared FPA detectors applied to a low-temperature background, the first aspect that must be confirmed in the design stage is whether the performance of the detectors at low temperatures meets the application requirements. Using the basic theoretical formula of the unit detector, from the perspective of testing, the main performance parameters of the theoretical calculation formula for FPA detectors were deduced, including the peak responsivity, noise, peak detectivity, and dynamic range. In addition, flowchart of the calculation design for the detectors was proposed. The main parameters of the long-wave infrared FPA detector at low temperatures were designed, calculated, and verified using an example, and the results are in good agreement with the theoretical calculations. Thus, the theoretical formula and design calculation flow are practical and can provide a reference for designing infrared FPA detectors.