Compared with n-on-p materials, p-on-n materials have lower dark current and higher operating temperature, and are more suitable for long wave and high temperature HgCdTe infrared focal plane devices. The research progress on long wave p-on-n devices by French Sofradir Company, American Raytheon Vision Systems Company, North China Research Institute of Electro-Optics and Kunming Institute of Physics is introduced.

Infrared detection system is widely used in aerospace, military reconnaissance and other fields due to its good concealment and strong penetration ability. However, the detection distance of the infrared detection system is far, and the target tends to be dim and small. The dim small target detection in infrared single frame images has always been a difficulty and research hotspot in the field of infrared detection. Based on the four aspects of filtering, visual saliency, image data structure and deep learning, the current single frame infrared dim target detection algorithms are reviewed in detail. Finally, the development of infrared dim small target detection technology is summarized and prospected.

The centering error of infrared multi-lens is one of the important factors affecting the imaging quality of the optical system. When the infrared detector works in the cryogenic environment, the pre-assembled and tested infrared multi-lens at room temperature will cause misalignment of assembly accuracy. A method to measure the centering error of infrared multi-lens in the cryogenic environment is proposed. The difficulty of measuring the low temperature position accuracy of lens group is solved by the designed low temperature test system. The test results show that the method can effectively measure the centering error of infrared multi-lens at low temperature, and the measurement error is better than 2 m. This research has certain practical significance for the development of high-performance infrared detector components.

In order to meet the application requirements of artificial intelligence in target recognition methods, the ability of intelligent classification, recognition and interpretation of massive data is necessary. The data in the infrared target characteristic database is further explored, and the infrared target recognition algorithm based on HOG+SVM is used in the infrared target recognition process. The collected targets such as cars, helicopters, planes, ships and unmanned aerial vehicles are selected, and the infrared target detection and classification algorithm is realized combining the HOG operator and the SVM classification method, so as to realize the intelligent infrared target classification research. It provides support for subsequent further analysis of the target characteristic and seeker intelligent algorithm design.

The nonuniformity of infrared detectors directly affects the quality of infrared imaging and measurement accuracy. Ground-based infrared radiation measurement systems often can not occupy the full surface of the detector when imaging long-distance flying targets. In order to improve image quality, a method of nonuniformity regional correction based on calibration is proposed. Taking the cooled mid-wave infrared detector with a target surface of 640×512 as the experimental object, based on the two-point calibration method of blackbody calibration, the whole surface calibration algorithm and the algorithm proposed in this paper are used for verification. The results show that when the imaging area is less than 1/3 of the surface, the nonuniformity error after region nonuniformity correction is less than 0.002%. Compared with the full-target nonuniformity correction algorithm, this correction algorithm further reduces the nonuniformity by about 30% to 75%, and the reduction rate of the nonuniformity error is greater than 30%. After using the correction algorithm in this paper, the nonuniformity of each area is further reduced, and the visual effect of correction is further improved. Therefore, the correction method has certain engineering application value.

Because the water environment capacity of river network in the integrated demonstration area of Yangtze River Delta is limited and the terrain is low, flood disasters usually appear in rainstorm season. Satellite remote sensing technology has become one of the important means of flood disaster monitoring because it is macroscopic and dynamic. In this paper, based on Sentinel-1 C-band synthetic aperture radar (SAR) images, the water body identification method of Dianshan Lake is studied, and the multi-temporal water area extraction is preliminarily realized during plum rain in 2020.Combined with FROM-GLC global land cover type data, the water accumulation situation is analyzed. According to the research results, the following conclusions can be drawn :(1) SNAP software of ESA has a good suppression effect on speckle noise of SAR images; (2) Since the backscattering coefficient of Dianshan Lake water body shows a single-peak shape on the image, the threshold segmentation method can be used to extract water information; (3) The water regime during plum rain increases by about 40km2 compared with that in April. (4) Under the influence of heavy rainfall, the area of submerged farmland is the largest, while impervious surface and bare land are less affected.