Abstract: Aimed at the problem that the particle filter algorithm is difficult to automatically initialize seed points in SAR image road extraction, a road extraction method of SAR image based on Hough transform and particle filter is proposed. The method took advantage of ROA operator to detect the edges, extracted the straight road line segments and road intersections in the image using Hough transform and morphological operation respectively, and regarded the endpoints of straight road line segments and road intersections as seed points to complete the subsequent global road tracking and connection according to the improved particle filtering model. The experimental results show that the method can effectively extract a variety of road shapes such as straight lines, curves and intersections without setting initialized particle filter seed points manually.