The algorithm research of infrared small target detection and recognition

This article researches the detection and identification of weak targets under the infrared undulate background, combining the actual test environment of shooting range based on the analysis of the existing image target detection and recognition algorithms limitations. The proposed real-time band-pass filtering algorithms, locally adaptive threshold fusion and segmentation algorithms based on statistical parameter adaptive computing and point polymerization, goal chain algorithms based on dynamic storage space allocation respectively detect probability from the low SNR infrared image, realize low SNR point, dot and surface target effective identification tracking under complex background and significantly reduce the amount of data storage that makes a useful attempt to improve detection and tracking in real-time. Test results show that the proposed algorithms have good recognition performances and robust adaptability for low SNR small infrared target under complex environment.