An object-oriented method for road damage detection from high resolution remote sensing images

摘要

The precision of road damage detection in disaster area from high resolution remote sensing images is general low because the current methods are mainly based on pixels and can't combine with the already existing GIS information. This paper presents an object-oriented road damage detection method. First and foremost, Multi-scale segmentation technology is adopted to get image objects, and then the paper utilizes some typical feature parameters and combines with pre-disaster's GIS vector road to extract road information. Then the techniques of overlay analysis and violating objects removal are employed to extract the damaged road section. Remote sensing images for Wenchuan, China disaster area and Yushu, China disaster area are implemented as examples. The result indicates that this method can improve quickness and efficiency of road damage detection.