Urban Land Use Classification from High-resolution SAR Images Based on Multi-scale Markov Random Field

Synthetic Aperture Radar (SAR) images are widely used in land use classification. However, the influences of buildings may cause classification errors when SAR images are applied in urban areas. Aiming at the property of low Signal to Noise Ratio (SNR) of SAR images and the complexity of building textures, we obtained the initial segmentation using the Maximum likelihood (ML) algorithm based on the multi-scale Markov Random Field (MRF) model and involved the Gabor similarity between pixels based on the traditional MRF potential function, and employed the Iterative Conditional Model algorithm to implement the segmentation. And we classified the segmentation image by using the K-means classification algorithm. The experimental results on several real SAR images showed that the proposed approach performs better than traditional methods in the segmentation accuracy, and building boundaries were clearly obtained by the proposed approach.