Non-Linear Diffusion Based Level Set Approach for Road Extraction Using High Resolution Imagery

Road extraction is one of the major problems in cartographic feature extraction. High-resolution satellite imagery provides an important data source for road extraction. This paper deals with road extraction from high resolution satellite imagery using level set method. In the proposed method, the high resolution satellite image is first preprocessed using non-liner anisotropic diffusion to enhance the road boundaries and to smooth the image. In the next step level set method is used to extract the roads or road network. The method is tested on high resolution satellite imagery such as IKONOS and Quickbird.