Abstract:

n order to solve the problem that the current approaches cannot suppress the background clutters effectively and result  in a poor detection performance, a novel infrared dim target detection approach based on fuzzy adaptive resonance theory  (Fuzzy-ART) neural network is presented. Firstly, the Fuzzy-ART neural network is combined with Robinson guard to build  the adaptive local spatial background models. With these models, the background clutters are suppressed according to the  degree of fuzzy match between pixels and models. Then a difference algorithm based on template average is utilized to  highlight the targets according to the spatial features of targets and residual background clutters. The proposed  adaptive segmentation algorithm based on fuzzy cluster of rows and columns is next used to detect the candidate targets.  Finally, the true targets are further detected by the multiframe trajectory related algorithm based on the consistency  of target motion. Theoretical analysis and experimental results show that the proposed approach can adjust the spatial  background models adaptively according to the condition of local background, and eliminate the background clutters and  highlight the targets effectively. It is capable of improving the signal-to-noise ratio and detecting the targets  effectively.