摘要
该文拟将计算机视觉中传统模板匹配方法与对场景物体的中心矩描述方法相结合，提出了一种效率较高的匹配算法。算法在匹配识别上不仅不受物体方位旋转及尺寸缩放变化影响，而且能较精确地求出物体相对于模板的缩放比例和偏转角度。文中给出了与传统模板匹配算法对照的实验结果。
A improved template matching algorithm of target recognition is proposed which combines invchable center moment description with traditional template matching method. The proposed algorithm can recognize objects much fast than the traditional algorithm and recognize objects correctly even the objects rotate or change scale. Experiment data are given also.
出处
《计算机工程与应用》
CSCD
北大核心
2000年第6期42-43,69,共3页
Computer Engineering and Applications
关键词
模板匹配
目标识别
算法
计算机视觉
图象处理
template matching, RST invariable center moment, algorithm of target recognition
