A robust SAR image geolocation accuracy evaluation method based on road intersection feature extraction

During the evaluation of SAR image geolocation accuracy, for different checkpoint types, different degrees of errors will be introduced when determining the pixel coordinate of the checkpoint on the SAR image by using the manual visual selection method, which reduces the reliability of the final evaluation result. Aiming at this problem, this paper proposes a robust SAR image geolocation accuracy evaluation method based on road intersection feature extraction. The center point of the road intersection is determined by extracting the straight line features of the road edges that constitute the road intersection and calculating the corresponding straight line intersections. Finally, two SAR images are evaluated to verify the effectiveness of the proposed method.