SAR image change detection method based on shearlet transform

Abstract

Multi-temporal synthetic aperture radar (SAR) images have been successfully used for the detection of different types of terrain changes. However, SAR image change detection based on wavelet transform is still restrained from the existence of speckle noise and the nature of wavelet transform. In this paper, an unsupervised SAR image change detection fusion framework based on shearlet transform is proposed. In the proposed method, The Gauss filtering is combined with log-ratio to impair speckle. Then the difference map (DM) of Gauss-log ratio and the difference map of ratio based on Gabor feature are fused with shearlet transform. Meanwhile, DM is decomposed to low frequency image and four high frequency images, different fusion rules are used in multi-scales images respectively, the work of noise reduction is operated with mean filtering. After an inverse shearlet transformation, the final change map can be obtained via a simple OSTU segmentation. The real SAR image pairs in Bern area are used to verify proposed change detection method. The experimental results demonstrate the robustness of the proposed method.