基于SAR影像平地与水域ROI的由粗到精匹配方法

A New Cursory-to-exacted Matching Algorithm Based on Region of Interest for Flat and Water Area

INS/SAR组合导航系统主要是基于图像相似性相关原理,即将SAR实时获取的图像与预存在匹配处理机中的基准图进行配准比较,得到载体相对于目标或预定区域的位置偏差,从而修正INS的累积误差,保证载体的导航精度。但由于基准图与实时图获取的条件不同,两者存在较大差异,会损害匹配概率和匹配精度,进而影响载体的定位精度。基于此提出了基于SAR影像平地与水域特征区域(ROI:Region of Interest)的由粗到精的匹配方法:即利用实时图全部小波第二层低频信息在基准图小波第二层特征影像上进行全局搜索,获取实时图中心点的粗匹配坐标,根据粗匹配坐标找到对应的基准图上的区域及所包含的平地与水域ROI,再找到ROI在实时图上对应的大致区域,利用互相关算法对ROI进行匹配,得到最终匹配坐标。并对该方法进行了仿真计算,结果表明该方法在星载SAR/机载SAR匹配中取得很好的效果,满足高匹配精度的同时,增强了不同源SAR匹配的适应性。

The work principle of the SAR/INS compound navigation system is based upon the correlation of image similarity, that is, getting the carrier position error faced to the target region or region through the image registration, then correctting the accumulative error of the INS and ensuring the navigation precision. Because of the biggish differentia between the real -time image and the base image that caused by the different imaging condition, the match probability and match precision decline, finally worsen the carrier position precision. A new cursory-to-exacted matching algorithm base on the fiat and water characteristic region （ROI） is put forward, that is firstly, getting the cursory match result through point by point research in the second scale of the wavelet transform, finding the rough region that contains the fiat area or water area in the base image based on the cursory result, then reversely, searching the ROI location （that is the final match results） in the real-time image using the cross-correlation measurement. The simulation results show that the matching algorithm is applied to the spaceborne SAR/airborne SAR successfully, not only increases the match precision, but also improves the robustness of the SAR registration.