Airport Runway Detection Based on Figure Segmentation and Classification in PolSAR Image

At present, the airport runway detection in Polarimetric Synthetic Aperture Radar (PolSAR) images is mainly based on classification methods, which is inefficient and requires more prior information. To solve this problem, an unsupervised runway detection method based on image segmentation and classification is proposed. Firstly, the pseudo scattering power (Pspan) of the image is defined, and image is thresholded by using Pspan and Tsallis entropy, so as to get the region of interest. Then, the classifier combining density peak search and Wishart distance is used for binary classification. Finally, according to the texture and geometric features of airport runway, the classification results are further identified to determine the real airport runway area. Experiments were carried out by using multiple groups of measured PolSAR data collected by different systems. The results show that the proposed method can detect all airports in the image, with clear outlines and complete structure, and the detection efficiency is improved.