A road extraction algorithm according to high⁃resolution synthetic aperture radar(SAR)image is proposed. The FROST filtering is conducted for SAR image to eliminate the coherent speckle noise. The Canny operator and ratio of average (ROA)operator are fused to extract the peripheral points,and compensate the information of the image edge. And then the pa⁃rallel filtering is used to filter out the influence of the related ground objects on road detection. The Hough transform is used to extract the line feature,and connect the broken lines. The experimental results show that the proposed algorithm can effectively filter out the influence of fence and other ground objects,and has good detection effect on straight road.%提出了一种高分辨率SAR图像的道路提取算法。首先对SAR图像进行FROST滤波，去除相干斑噪声，再融合Canny算子和ROA算子提取边缘点，并对图像边沿的信息进行补偿，然后使用平行线滤波滤除相关地物对道路检测的影响，最后利用Hough变换提取线特征，并对断裂的直线进行连接。实验结果表明，该方法能有效滤除护栏等地物的影响，对直线道路具有良好的检测效果。
展开▼
