Damaged Road Extraction From Single-Temporal Semi-Post High Spatial Resolution Remote Sensing Imagery

After an earthquake,roads act as important transportation junctions and facilitate post-disaster rescue and transportation of goods.It is critical to efficiently identify semi-post damaged roads.In this paper,an extraction method for semi-post damaged roads is established,based on single-temporal high resolution remote sensing imagery,and the method can resolve the problem that data is shortage when based on multi-temporal high resolution remote sensing imagery.In this method,the image is initially multi-scale segmented and classified using the nearest neighbor approach,the resulting classification map is re-classified into two classes,the small and the large patterns in the binary classification map are morphologically removed,and finally,Hough transform is applied to extract damaged roads and calculate the road lengths.In experiments using WorldView-2 imagery of the earthquake occurred in Ludian County,Yunnan Province and using UAV imagery of the earthquake occurred in Beichuan County,Sichuan Province,road extraction is done on the basis of object-oriented classification,in order to illustrate the superiority of object-oriented classification,the proposed method is compared with a pixel-based classification method,and the accuracies of the two methods are quantitatively analyzed by using confusion matrix.Based on the extracted semi-post damaged roads,the accuracy and damage type of the proposed are evaluated with reference to visual interpretation and a road damage assessment model.It is demonstrated that the proposed ethod can make the precision more than 80%and efficiently offer damaged roads for earthquake rescue.