Earthquake-damaged buildings detection using active contour algorithm and MRF

As damaged buildings after earthquake cannot be efficiently detected only based on remotely sensed image, a novel method is proposed for detecting damaged buildings. This approach combines improved active contour algorithm and Markov Random Field (MRF). Firstly, the building regions are identified according to pre-event GIS data, and the improved active contour algorithm is used to extract homogeneous regions in building boundaries from the post remotely sensed image. Secondly, easily identified buildings are detected by shape similarities between extracted and initial shapes with double thresholds. Thirdly, for the uncertain buildings between two thresholds, the improved MRF with distance weight is utilized to build spatial relation model among buildings accurately to determine whether they are damaged. The experimental results indicate that the proposed method can efficiently improve the accuracy and robustness to detect earthquake-damaged buildings by remotely sensed images. An effective way is provided for detecting damaged buildings and emergency relief after an earthquake.