Due to the characteristics of small size and light weight of unmanned aerial vehicles (UAVs), wind has great influence on UAV flight, and images photographed by UAV suffer large distortion. Therefore, it has practical significance to correct the distortions before image matching. We focus on extracting the edge contour based on Canny operator and obtain the large number of surface features contour using Hough transform theory. Then step approximation is used to obtain inverse solution model of the real image pixel coordinates, which is used to correct the results. Meanwhile, on the foundation of parallel and perpendicular terrain contour lines, the correction methods of quadrilateral connection and parallel lines constraint are proposed. It is found through experiments that the achieved results are satisfactory.