Improved self-adaptive negative selection algorithm with double clustering for infrared target extraction

Since there are a large number of holes, which cannot be detected by Negative Selection Algorithm (NSA), this paper proposes a self-adaptive selection algorithm with double clustering. In the first clustering, OTSU is used to calculate the optimal threshold. According to the Hamming distance matching rule, the self-covering detectors and non-self-covering detectors are formed. In the second clustering, the detectors are used to extract the target. Experimental results show that the improved algorithm can achieve a wider range of non-self space coverage, improve the detection rate of holes, and obtain more accurate detection performance.