基于遗传算法的快速图像相关匹配

红外与毫米波学报990210

红外与毫米波学报

JOURNAL OF

INFRARED AND

MILLIMETER

WAVES

1999年　第2期　No.2

　1999

基于遗传算法的快速图像相关匹配*

朱　红　赵亦工

摘要　采用遗传算法研究了图像相关匹配问题,提出了快速图像相关匹配算法在最优匹配的前提下,其计算量较SSDA算法降低了一个数量级以上;可以采用NPROD匹配准则,改善在低图像对比度条件下的匹配精度;该算法每帧图像匹配计算时间基本恒定,便于工程应用.

关键词　图像相关匹配，遗传算法，快速计算.

FAST IMAGE CORRELATIVE MATCHING BASED ON

GENETIC ALGORITHM

ZHU Hong　ZHAO Yi-Gong

(Department of Measurement-control and Equipment, Xidian University,

Xi’an, Shaanxi 710071, China)

Abstract　The problem of image correlative matching was studied by means of genetic algorithm and a fast image correlative matching algorithm was presented. With the condition of optimum matching, the computation amount of the algorithm was reduced more than 10 times compared with SSDA algorithm. The NPROD matching criterion can be employed in this algorithm to improve the matching precision for the case of low image contrast. The computation time is nearly constant for matching each frame of images in the presented algorithm, which brings convenience to the practical application.

Key words　image correlation matching, genetic algorithm, fast computation.

引言

图像相关匹配跟踪技术是光电成像系统对运动和静止面目标跟踪的基本手段.对于近距离面目标或尺寸很大的面目标,目标图像占据视场的大部分或充满视场,图像相关匹配的数据量和计算量很大.图像相关匹配的计算实时性在一定程度上决定了该技术的实file:///E|/qk/hwyhmb/hwyh99/hwyh9902/990210.htm（第 1／8 页）2010-3-22 23:46:14