A Novel Method to Solve the Shrinkage Problem of Targets in Infrared Small Target Detection

Infrared small target detection methods based on the local contrast have attracted much attention. For some larger targets with clearer outlines, along the direction of the target center pixel to the edge pixel, the contrast calculated using the sliding window will gradually decrease to a value that tends to zero, resulting in a shrinkage of the target in the final saliency map, which fails to achieve the expected detection effect. For this reason, this paper proposes a novel method based on local contrast to solve it. First, a strategy to suppress the background based on multi-directional gray-scale jumps is proposed after analyzing the gradient characteristics of the target and the edge clutter. A method is designed to generate a zero region of the target, which can make the pixel value of the neighborhood background of the target in the image where the background is suppressed to zero. Then the contrast is calculated in the zero region using the nested protection window mechanism, which can expand the area of the real target. Finally, the target is re-extracted from the enlarged target area by using the perceptual prior obtained by the mean filter. Extensive experimental results show that the method proposed in this paper has high detection performance.