LAUSR

In general, changes in the multitemporal synthetic aperture radar (SAR) images are detected by classifying the SAR ratio images into the changed and unchanged classes. However, multitemporal SAR images have either increase or decrease in the backscattering values. Therefore, the changed areas can be further classified into positive and negative changed classes. This paper presents an unsupervised change detection approach for detecting the positive and negative changes based on a multi-objective evolutionary algorithm. In this paper, the widely adopted mean-ratio and log-ratio operators are extended to generate SAR ratio images for distinguishing the positive and negative changes. In order to reduce the corruption of speckle noise present in the multitemporal SAR images, a fuzzy cluster validity index is established to exploit local spatial and gray level information. Then the objective functions are simultaneously optimized by a multi-objective evolutionary algorithm. The experimental results on two simulated data sets and three real SAR data sets confirm the effectiveness of the proposed method.