Abstract: The edge pixels of airport runway may be misclassified due to category ambiguity. Therefore, a runway area detection method for PolSAR image based on robust fuzzy C-means clustering is proposed. Firstly, in order to avoid the sensibility to initial centers and local optimum occurring in fuzzy C-means clustering, the polarization decomposition theory is utilized to segment the original image and obtain appropriate initial centers. Then the regions of interest (i.e. suspected runway areas) are extracted from the classification results, which are obtained by the robust fuzzy C-means clustering method based on Wishart distance. Finally, structural features of runway are used to identify the extracted areas and finalize the real runway area. Simulation results show that the proposed method has higher accuracy, faster speed, and better visual effect comparing with the other two algorithms.

Key words: polarimetric synthetic aperture radar, runway area detection, fuzzy clustering, Wishart distance, structural characteristics