Automatic road extraction based on improved radon transform

In this paper, a new method is proposed for road extraction in complex environments. The algorithm takes full advantages of the road image characteristics. Firstly, the image is preprocessed and the object edges are extracted. Then, by dividing the image into many blocks (such as m×m), an improved Radon transform (RT) is performed to extract the line segments in images. We do Radon transformation for each block. To reduce the computing time, a threshold is set and images are resumed along the ribbon region of transformation domain. Experiments show that the presented method can extract roads even in complex environments, and moreover, it can provide a complete description of the road.