Background Suppression Algorithm for Infrared Images Based on Robinson Guard Filter

Given the limitations of traditional Robinson guard filter, a novel algorithm is proposed to suppress background of infrared (IR) images. First, according to human visual contrast mechanism, regions of interest (ROIs) can be obtained. Then approximate sizes of candidate targets can be acquired by extracting outlines of ROIs. Based on that, the window size of Robinson guard filter can be adjusted adaptively. Finally, this paper adopts multi-scales to optimize further to obtain the best filter window size. This algorithm solves the limitations of traditional Robinson guard filter. The experimental results demonstrate that the proposed algorithm is adaptive to small targets with different sizes. It also has good ability to suppress background and enhance small targets for low SNR(signal-noise ratio) images.