Road Extraction from High-spatial-resolution Remote Sensing Image by Combining GVF Snake with Salient Features

The road information in the high-spatial-resolution remote sensing image is of great significance for updating the GIS database. Through analyzing the road features shown in the remote sensing image, this paper presents a road detection method based on salient features and gradient vector flow (GVF) Snake. According to the visual cognition theory, the road geometric and radiation features are viewed as salient features in this paper. First, the saliency map is calculated by fusing the color-based and structure-based contrasts, the maximum saliency value is regarded as the seeds of the GVF Snake. Then, a region-growing algorithm is applied to compute the initial boundaries, the energy function of the GVF Snake is minimized by iterative solution of the gradient vector flow model to get the final road information.Experimental results show that the proposed method could enhance the computational efficiency and has good detection accuracy