快速并行两级景象匹配方法

文章分析了景象匹配系统的性能评价以及影响性能评价的三个主要因素,即实时图大小,参考图大小和匹配算法。分析了实时图大小的选取对匹配时间,匹配率的影响。并且针对巡航导弹的景象匹配制导系统的实时性问题,提出了一种快速并行两级景象匹配方法。该方法能够极大地提高匹配效率。试验中利用SAR图像和可见光图像进行景象匹配,每次匹配的时间从原来的约16秒减少到300毫秒左右,匹配率可达到90%以上。而且还具有可扩充性好的特点。

An Efficient Parallel Algorithm of Two-Phase Image Matching

This paper gives performance estimate of image matching system.It uses matching time ,ratio and error to judge behaviors of a system.There are three mainly factors that have effect on the system' performance.They are size of real time image,size of reference image and efficiency of matching algorithm.In order to choose appropriate size of real time image,it presents some relations between real time image's size and other factors.To use image matching in cruise missile,a fast parallel algorithm of two -phase image matching is proposed.By two -phase processing and parallel method,experiments using SAR image and optical image show that the new approach can reduce matching time from almost sixteen seconds to about three hundred milliseconds while matching ratio is above90percent and error less than four pixels.The algorithm works quite accurately and efficiently.