Infrared Small Target Detection Based on the Effective Rank-Layering of Image

The low rank decomposition algorithm in the field of infrared small target detection has the problem that it is difficult to distinguish between small targets and clutter edges. In this paper, an efficient small target detection algorithm based on image effective rank layering model is proposed, which can effectively solve this problem. Implementing this method requires two steps. In the first step, the original infrared image is rearranged into operational matrix using local patch construction, which can improve the non-local self-correlation property of the infrared background image. The inexact augmented Lagrange multiplier (IALM) algorithm is then used to process the operational matrix to recover the background and target foreground images. In the second step, effective rank-layering is used to process the target foreground image. Here, the target foreground matrix is first subjected to singular value decomposition, and then the singular value attenuation is applied to determine the effective rank of the matrix, which represents the number of layers in the image. Next, using the minimum image information entropy to select the image layer in which the infrared small target is actually located. Finally, a threshold image segmentation technique is used to refine the detection result. Experimental results demonstrate that the proposed method outperforms the comparison algorithms, and it can effectively suppress the clutter background under different complex detection backgrounds.