基于权值模板匹配算法的全自主足球机器人目标识别  

Target Recognition of an Autonomous Robot Soccer Based on Weight Template Matching Algorithm

作　　者：吴晓[1,2] 曹其新[1] 

机构地区：[1]上海交通大学机械与动力工程学院,上海200240 [2]莆田学院电子信息工程学系

出　　处：《厦门大学学报（自然科学版）》2008年第6期818-822,共5页Journal of Xiamen University：Natural Science

基　　金：国家自然科学基金(60443007)资助

摘　　要：提出了一种基于权值模板匹配的全自主足球机器人目标识别方法.首先通过动态窗口、权值变换、图像索引和动态模板等方法提高了目标的识别率,降低了识别时间;然后在足球机器人视觉系统中进行了仿真实验和实地测试.实验结果表明,在原图像含有30%白噪声情况下,其识别率仍能达到99.7%,且识别时间不超过30 ms,这种方法既在一定程度上降低了噪声干扰的影响,又提高了目标识别的速度和准确性,从而提高了系统的实时性和鲁棒性.This paper proposes an autonomous robot soccer target recognition algorithm based on weight template matching. It improved the target recognition rate and decreased the recognition duration through method such as dynamic window, weight conversion,image index and dynamic template. Besides that,it carried out simulation experiment and field testing on the vision system of robot soccer. Result of the experiment shows that when the original image contains 30% of white noise,its recognition rate can still reach 99.7% and the recognition duration is no more than 30 ms. This method not only decreased the effect of noise interference to a certain extent,but also increased the speed and accuracy of target recognition,thus increased the real-time and robustness of the system.

关 键 词：模板匹配 全自主足球机器人 权值算法 目标识别 

分 类 号：TP242.62[自动化与计算机技术—检测技术与自动化装置] TS251.52[自动化与计算机技术—控制科学与工程]