景象匹配辅助导航中多级实时匹配算法研究

由于选用部分Hausdorff距离进行图像匹配时不能满足较大尺寸图像匹配实时性要求,利用了多级分层匹配技术加速图像匹配,提出了多级实时图像匹配算法,该算法以部分Hausdorff距离为图像匹配相似性度量,结合小波变换多分辨率思想提取多尺度图像特征,实现了由粗到精多级分层实时匹配。图像匹配结果表明,光学卫星(IKONOS)图像和合成孔径雷达(SAR)图像二种情况下图像匹配结果准确,比仅采用基于部分Hausdorff距离图像匹配而言,采用三级图像匹配可大大缩短匹配时间。该算法实施景象匹配能在5 s以内完成,满足景象匹配辅助导航系统匹配修正的准确性和实时性的要求。

Study of hierarchical real-time matching algorithm for scene matching aided navigation system

The part Hausdorff distance can't meet the real-time performance requirements for large size images matching,so the hierarchical matching technology has been utilized to speed the image matching,and a hierarchical real-time image matching algorithm has been given.Based on the part Hausdorff distance,which is chosen as the similarity measuring during image matching process,it has utilized the multi-resolution analysis of wavelet transform for extraction of image multi-scale features,and realized the hierarchical real-time matching from roughness to precision.The results of image matching show that is accuracy in all of optical satellite(IKONOS) and synthetic aperture radar(SAR) images,and contrast to the part Hausdorff distance image matching,the use of the three levels image matching can quickly shorten the matching computing time.The scene matching based on the algorithm can be finished within five seconds,so it can meet the accuracy and real-time performance requirements for matching modified in the scene matching aided navigation system.