摘要
提出一种自适应选取阈值的三层快速匹配算法。采用了由粗到精的匹配策略,先分别用环形模板和十字形模板进行粗匹配来淘汰大量非匹配点,再用全模板进行精匹配确定匹配位置以减小计算量。以模板和实测图的平均灰度差来自适应选取阈值,增强了算法的稳定性,从而弥补了环形法的模板信息单一且阈值固定的缺点。实验结果表明,该算法的匹配位置准确,匹配速度比相关法提高5倍,比环形法提高10倍,且在目标发生明显运动变化时仍具有很强的可靠性。
A three-layer fast matching algorithm based on adaptive selection of threshold is proposed. A strategy from coarse matching to fine matching is adopted. Large numbers of non-matching points will be firstly eliminated by performing coarse matching with circular and cross templates and then a whole template is applied to confirming the final position to reduce the calculation workload. The threshold is adaptively selected by the difference between the mean gray scale of the template and the object image for the purpose of improving the stability of the algorithm, thus the shortages of the algorithm using circular template that the information is limited and the threshold is fixed, can be compensated. Experimental results show that the matching position of the algorithm is correct, its matching speed is 5 times higher than that of correlation method and 10 times than that of circular method. It still has a high reliability when a target moves obviously.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2005年第6期83-85,96,共4页
Opto-Electronic Engineering
关键词
图像匹配
目标跟踪
自适应阈值
环形模板
十字形模板
Image matching
Target tracking
Adaptive threshold
Circular template
Crisscross template
