A Novel Road Extraction Algorithm for High Resolution Remote Sensing Images

Automatic road extraction from the high resolution remote sensing images is of great importance in intelligent transportation and image processing. Hence, in this paper, an effective road extraction algorithm for high resolution remote sensing images based on the circular projection transformation is proposed. The main idea of the proposed algorithm lies in that the road extraction results are obtained by selecting a suitable initial template, and then search the matched templates through moving the initial template in two directions. Firstly, the circular projection vector of the initial template is achieved by calculating the circular projection value at a specific radius. Secondly, the optimal radius of the circle in circular projection transformation and the length of the seeking step and the seeking angle are determined. Thirdly, for each seeking step the similarity between the target template and the initial template is computed, and the template with the highest similarity is chosen. Finally, roads can be detected by the correct direction by exchanging the first two detected points. To make performance evaluation, the IKONOS dataset is utilized and DMES and AUA algorithm are compared. The experimental results demonstrate that the proposed algorithm can automatic the roads from high resolution remote sensing images effectively and efficiently.