Unsupervised Multitemporal Building Change Detection Framework Based on Cosegmentation Using Time-Series SAR

Building change detection using remote sensing images is essential for various applications such as urban management and marketing planning. However, most approaches can only detect the intensity or type of change. The aim this study to dig more information from time-series synthetic aperture radar (SAR) images, frequency moments. This paper proposes a novel multitemporal building framework that generate map (CFM) moment maps (CMMs) SAR images. We first give definitions CFM CMMs. Then we feature four proposed generators. After that, new cosegmentation method combining raw divide into changed unchanged areas separately. Secondly, morphological index (MBI) are combined extract objects. Then, logical conjunction between results binarized MBI performed recognize every In post-processing step, use fragment removal increase accuracy. Finally, propose accuracy assessment CFM. call average difference (ACD). Compared traditional methods, our outperforms other in terms both qualitative quantitative indices ACD two TerraSAR-X datasets. experiments show effective generating