基于SAR图像的快速景象匹配方法

摘要

As an important part of the compound guidance,scene matching guidance requires the matc-hing algorithm with low computational burden and high matching accuracy.Focusing on this problem,a new scene matching algorithm based on normalized cross correlation and improved partial Hausdorff distance is proposed.In order to alleviate the computational burden,the image edge is used as feature space,and the original image is decomposed into a series of sub-images by wavelet transform.Therefore,the scene matc-hing is translated into the sub-image matching.Meanwhile,in order to enhance the matching accuracy,the sub-images are matched by normalized cross correlation method coarsely at first,and then the improved par-tial Hausdorff distance is applied to obtain the final matching point.The experiment result shows that this al-gorithm owns less computational burden and better matching accuracy compared to the traditional methods.%景象匹配制导作为复合制导中的一种重要制导方式，其要求匹配算法在保证匹配时间短的同时具有较高的匹配精度。针对这一问题，提出了一种归一化互相关与改进的部分 Hausdorff距离复合的景象匹配算法。为了降低匹配时间，该算法选取图像边缘为特征空间，采用小波变换将原始图像分解为一系列维数较小的子图像，进而在子图像上逐层进行匹配；同时为了提高匹配精度，在子图像上采用归一化互相关算法进行粗匹配，然后在原图上粗匹配点的邻域内利用改进的部分 Hausdorff距离完成精匹配，获得精确的匹配位置。仿真结果表明，与传统算法相比，该算法具有较短的匹配时间与较高的匹配精度。