基于动态模板匹配的移动机器人目标识别  

Mobile Robot Target Recognition Based on Dynamic Template Matching

作　　者：彭玉青[1] 李木[1] 高晴晴[1] 张媛媛[1] 

机构地区：[1]河北工业大学计算机科学与软件学院,天津300401

出　　处：《传感技术学报》2016年第1期58-63,共6页Chinese Journal of Sensors and Actuators

基　　金：天津市自然科学基金项目(13JCYBJC15400)

摘　　要：将视觉显著性与基于动态模板匹配的目标识别方法相结合,提出了一种适用于未知环境下的移动机器人目标识别方法。具体而言,首先设计了基于分布式控制的移动机器人视觉系统,提高了视频处理效率;之后利用基于背景先验的显著性检测方法对图像进行预处理,排除了相对次要的背景区域;最后对处理后的图像进行动态模板匹配,提高了目标识别的准确率。实验结果表明,该方法能够满足移动机器人在目标识别过程中对图像处理的实时性和准确性的要求,具有良好的有效性。A target recognition strategy under unknown environments for mobile robot was presented with the combi-nation of vision salience and the target recognition method based on dynamic template matching. Specifically,a mo-bile robot vision system based on distributed control was designed for improving the efficiency of video processing.Secondly,the saliency measure based on background priors was used for image preprocessing and excluded thebackground regions which were less important. Finally,the recognition accuracy of the target was improved accord-ing to the dynamic template matching to images which had been processed. The experimental results show that themethod can meet the requirements of real-time and veracity in the process of mobile robot＇s target recognition,themethod show good effectiveness.

关 键 词：移动机器人 目标识别 模板匹配 视觉显著性 分布式控制 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]