“…Therefore, most algorithms for small target detection in infrared imagery, in general, deal with two steps, i.e., background suppression or clutter removal, and detection. Usual background suppression methods are mainly categorized into the following: 1) filtering methods, such as template filter [1], morphological filter [2], [3], high pass filter [4], wavelet based filter [5], etc; and 2) statistical learning methods [6][7][8][9], such as parametric learning [6], facet model fitting [7], nonparametric regression [8], etc. The filter based detection methods basically regard small target as an anomaly point in image scenes and separate target from clutter background, which rely on the differences between background and target in properties.…”