Enhanced GP-PNF Detector for PolSAR Image Pixel-level Detection using Target Characteristics Modeling and Superpixel Segmentation

A polarimetric synthetic aperture radar (PolSAR) target detection method based on geometric perturbation filter and superpixel segmentation is proposed to address the issues of high computational complexity and insufficient utilization of polarimetric information in existing pixel-based and superpixel-based target detection methods. In the detection stage, the PolSAR image is first decomposed into polarimetric components, and the polarization characteristics of the target are modeled. Then, a set of geometrical perturbation-polarimetric notch filter groups (GP-PNF) is constructed to enhance the contrast between the target and clutter and achieve coarse localization of the interested targets in complex scenes. The detection results of GP-PNF are further subjected to superpixel segmentation, and a superpixel-level polarimetric dissimilarity metric (SCDM) is defined to distinguish further the polarization characteristic differences between targets and clutter in local regions, reducing false alarm rates. Comparative experiments were conducted using real measured polarimetric data of ground background with various classical detection algorithms. The results demonstrate that the proposed method can effectively detect targets in complex backgrounds of PolSAR images without requiring background modeling.