Infrared small target detection based on clustering idea

In order to solve the problem of detecting infrared small targets of unknown size in complex background, an infrared small target detection algorithm based on the clustering idea is proposed. First, the original infrared image is preprocessed by using small target morphological features to generate a new density feature map. Secondly, the potential candidate targets are coarsely localized with an improved density-peak clustering algorithm. Then, the local candidate sets of potential targets are constructed. A weighted fuzzy set clustering algorithm is used to finely segment the target and background regions of the image block, and then the difference between the target and background is adopted to suppress false alarms while enhancing the target. Finally, an adaptive threshold is applied to the processed local candidate set to extract the real target. Experimental results show that the proposed algorithm has good robustness and detection performance for small targets of unknown size in comparison with the other seven methods.