摘要
针对现有模板匹配算法对运动目标物体的误判率高的问题,在已有模板匹配算法基础上,提出了一种权重融合动态模板匹配算法。设计了双目检测试验平台,阐述了模板匹配算法的理论基础,在归一化平方差匹配方法、归一化相关匹配方法和归一化相关系数匹配法基础上,提出了权重融合匹配算法。即给3种归一化的匹配方法分配不同的权重,误判率相对较低的分配的权重最高,误判率相对较高的分配的权重最低。在所设计试验平台上,在300m以内对运动目标进行了识别与跟踪。实验结果表明,该算法可以提高双目视觉系统对运动目标识别的准确性。
Aiming at the problem that the current template matching algorithm has high false positive rate for the moving target object,a weight fusion dynamic template matching algorithm was proposed based on the existing template.A new binocular detecting and testing platform was designed and the theoretical basis of the template matching algorithm was expounded.A weight fusion matching algorithm was put forward based on the normalized square difference matching method,the normalized correlation matching method,and the normalized correlation coefficient matching method,of which,the weights were different：the weights of the relatively low misclassification were the highest,while those of the high misclassification rate were the lowest.The moving target was identified and tracked within 300 meters on the designed test platform.The experimental results showed that the algorithm could improve the recognition accuracy of binocular vision system for moving target.
出处
《机械与电子》
2017年第1期77-80,共4页
Machinery & Electronics
关键词
动态模板匹配
双目检测
权重融合匹配算法
目标识别
dynamic template matching
binocular detection
weight fusion matching algorithm
target recognition
