Line segment detection of urban area in SAR images based on improved gray Hough transform

As one of important applications in Synthetic Aperture Radar (SAR) images, the recognition of urban area has received considerable attentions in remote sensing. The extraction of line segment is very critical technology to recognize the urban area because many objects such as streets and buildings are line segment. The common method to extract line segment is Hough transform, but most of the previous methods are based on binary images. So we have to select a threshold to binarizate the image, but at most time we can not determine the threshold properly, resulting in the lost of useful information. To solve the problem, an improved Hough transform algorithm on gray level, which can make the extraction of line segment independent of the noise and the length of line segment, is proposed. The approach is validated by the analysis of SAR images.