An adaptive multiscale approach to unsupervised change detection in multitemporal SAR images

A novel adaptive multiscale approach to unsupervised change detection in multitemporal synthetic aperture radar (SAR) images is proposed. This approach is based on a multiresolution decomposition of the log-ratio image (obtained by a comparison of a pair of co-registered images acquired at different times on the same area) in a set of scale-dependent images characterized by a different trade-off between speckle reduction and preservation of geometrical details. For each pixel to be analyzed, a sub-set of reliable scales is identified according to an automatic local analysis of the statistic of the data. The final change-detection map is obtained according to an adaptive scale-driven fusion algorithm, which properly exploits the results of the analysis at different scales for producing an accurate and reliable change-detection map in both homogeneous and border areas. Experimental results confirm the effectiveness of the proposed technique.