Abstract: A change detection method combining Nonsubsampled Contourlet Transform (NSCT), saliency detection, and difference map fusion was proposed to solve the problem of unstable accuracy in SAR (Synthetic Aperture Radar) image change detection in traditional methods. Firstly, SAR images were pre-treated by Frost filters and two types of difference maps for preprocessed images with mean ratio and domain logarithmic ratio being constructed. Secondly, the two types of difference maps were decomposed into low-frequency images and high-frequency sub band images through NSCT transformation, and SR algorithm was used to extract salient information of high-frequency sub bands. Then the adaptive fuzzy logic algorithm and the maximum modulus algorithm were used for image fusion to obtain the NSCT inverse transform image. Finally, the SAR change detection result map and evaluation data were obtained through the k-means clustering algorithm. The experimental results show that the proposed method, compared with traditional methods, has higher stability in SAR image change detection, with an accuracy ranging from 0.65 to 0.8, verifying the effectiveness of the method.

Key words: NSCT, SAR images, SR algorithm, image fusion, image change detection