Detection of Roads from Images Based on Edge Segmentation and Morphological Operations

Extraction of roads from high resolution satellite images has an important role in such tasks as urban planning, traffic management, navigation, map updating and etc. This paper presents an automatic method for roads extraction from satellite images. The proposed approach uses the method of edge segmentation on the bases of two-dimensional Markov chains. The original image is converted to the Lab color space, and the B component is used for edge detection. We use colour feature and threshold processing to separate the resulting segments into roads and backgrounds. To improve the quality of road extraction, the filtering by region size, skeletonization and morphological operations are used at the post-processing stage. Experimental results show the effectiveness of the proposed approach.