An improved robust estimation algorithm for small IR target detection

Directed against the problem of detecting small infrared target in complex background, this paper presents an adaptive background prediction algorithm based on improved robust M-estimation. In the proposed algorithm, target pixels and observed noise form the mixed interference of background estimation. In order to better estimate background, the authors of this paper introduce a correction factor to reduce the influence of target pixels on background estimation and a forget factor to make it more adaptive to estimation of complex non-homogeneous backgrounds. Experimental results show that the proposed algorithm can better suppress background and preserve target information than the commonly-used median filtering algorithm and LMS filtering algorithm. As a result, the proposed algorithm can detect effectively IR target in complex backgrounds and obtain great robustness.