基于计算机视觉反馈的AGV定位停车研究  

AGV Positioning and Parking Based on Computer Vision Feedback

作　　者：魏明明[1] 汪焰恩[2] 

机构地区：[1]西安工业大学机电工程学院,陕西西安710032 [2]西北工业大学机电学院,陕西西安710072

出　　处：《机床与液压》2011年第15期35-37,51,共4页Machine Tool & Hydraulics

基　　金：陕西省教育厅专项科研计划项目(09JK484)

摘　　要：为了实现AGV的准确定位停车,采用视觉反馈法。首先经过图像处理,当发现定位停车标志符后,小车记录此时标志中心点到车辆中心的距离,并根据直线运动公式对距离进行判断,当距离为零时,小车停止。在标志符识别及配准过程中,采用了统计特征区域黑点数与模板图中的黑点数对比来筛选模板的方法以减少模板匹配的计算量,并利用SSDA算法进行图像配准。结果表明:在标识符形状差别较大、黑点数距离比较远的情况下,点的匹配率达到98%,识别准确率较高。In order to achieve AGV positioning and parking accurately, this method based on computer vision feedback was used. When detecting the symbol of AGV positioning and parking, which was processed firstly, the center distance of AGV to the symbol was recorded. The distance was determined by AGV based on formula of linear motion. When the distance was zero, AGV stopped. In the process of symbol recognition and registration, in order to reduce the computational workload of template matching, the model was filtered by counting the black point number in the character area and contrasting with that of the model. The image registration was achieved using SSDA arithmetic. The results show that the matching rate is above 98% and recognition accuracy is high, under the condition of large differences in the symbol shape and the large distances of black point number.

关 键 词：自动导引小车 定位停车 标识符 识别 模板匹配 图像配准 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]