Robust detection of SAR/IR targets via invariance

One of the most challenging problems in automatic target recognition is reliable detection of targets in high clutter backgrounds. When the clutter statistics are unknown or highly variable, the false alarm rate of classical detection algorithms, e.g. the matched filter, cannot be controlled and target detection become unreliable. The reason for this is lack of robustness of the test statistics to clutter variations. We apply maximal invariants to design target detection algorithms which have constant false alarm rate yet maintain high target detection rate. Numerical comparisons are presented for multispectral and multisnapshot radar images which illustrate that significant gains are achievable using invariance approaches.