Automatic Detection of Airport Runway Area Based on Super-Pixel PolSAR Image Classification

摘要

This paper proposes an unsupervised algorithm for airport runway area detection based on super-pixel PolSAR image classification. First, the simple linear iterative clustering (SLIC) algorithm are used to obtain super-pixel image by segmenting the PauliRGB image in order to reduce computational complexity and save computing time. Then, VAT-DBE algorithm is used to estimate and obtain the number of clusters of the image automatically. Combing the polarization information, the super-pixel image is classified by the method of spectral clustering. After that, the suspected airport runway area is extracted according to the scattering characteristics of the runway and classification result. Finally, the airport runway area is detected by using structural and topological characteristics of the runways. The experimental results show that the proposed algorithm can detect the airport runway area effectively with a clear outline, complete structure, and low false alarm rate. It also needs less time and a priori information compared with other methods.