Abstract: Aiming to resolve the problem that the complex background of sea-sky and also pixel-level noise are easy to result in false alarm in the process of target detection, a detection algorithm of the infrared weak target using visual contrast mechanism is proposed. First, a contrast-enhanced image is obtained by using the defined local contrast measure operator. This step can enhance the visual saliency of the target region, and simultaneously suppress the interference of the complex background and pixel-level noise, so as to improve the signal-to-clutter ratio (SCR) of the image. Then, the saliency region of the image is optimized in multi-scale to improve the versatility of the algorithm, so that it can be competent in the detection of weak targets of different sizes. Finally, an adaptive threshold segmentation is used to obtain the real target. The experimental results show that the proposed algorithm can realize the robustness detection of different sized weak targets without image preprocessing. Thus it is an effective method for infrared weak target detection compared with other algorithms with its high rapidity, efficiency and strong applicability.

Key words: infrared weak target, visual contrast mechanism, local contrast measure, multi-scale, threshold segmentation