Simple Roads Extraction Algorithm Based on Edge Detection Using Satellite Images

This paper presents a novel algorithm for extracting the road networks based on Canny edge dectector. The technique commences with the pre-proessing stage where the Contrast Limited Adaptive Histograph Equalization is implemented. Next, the image enters the filtering stage and the Wiener filter is applied to reduce the noise level. Then, the edge detection is carried out by applying the Canny detector. After that, the image is elongated with further supressing the noise at the borders. Finally, a mask created from the filtering stage with the dilated image are used to construct the roads. To test this technique, high spatial resolution images from Dubai- Sat are utilized. The attained results give maximum detection rate 98.77% This result is the best reported of its kind in the literature. So far, the edge detection is used to identify the lanes. However, no research has been reported to extract the road based on edge detection. Moreover, most reported alorigrithms extract the centerline or the skeleton of the roads not the full structure. Thus, this technique brings new insight to road extarction algorithms. Since it employs the edge detection, it does not require to identify the starting pixel and it can recognize both striaght and curvilinear roads.