视觉机器人货物轮廓提取与定位  

Goods contour detection and positioning by a vision robot

作　　者：丛奎荣[1] 韩杰[1] 常发亮[1] 

机构地区：[1]山东大学控制科学与工程学院,山东济南250061

出　　处：《山东大学学报（工学版）》2010年第1期15-18,共4页Journal of Shandong University（Engineering Science）

基　　金：国家自然科学基金资助项目(60775023;60975025);山东省自然科学基金资助项目(Z2005G03);山东省博士基金资助项目(2006BS01012)

摘　　要：提出一种视觉机器人货物拣选中的货物轮廓提取与定位方法.在图像预处理的基础上,采用Sobel算子提取货物与货箱边缘图像I1[i,j],通过图像投影取得货物支持区及其边缘图像I2[i,j],对货物支持区利用Canny算子提取边缘图像I3[i,j],利用I3[i,j]对I2[i,j]进行边缘连接,最后边缘跟踪提取货物轮廓实现货物定位,控制机械手运动并实现货物抓取.实验结果表明该方法实用有效.A new method of goods contour detection and positioning by a vision robot was proposed. Based on image pre-processing and the Sobel operator, the edge of goods and its container I1 [ i, j ] was detected. The goods-supporting area and its edge image I2 [ i,j] were calculated＇by image projection. Then another goods-supporting area＇s edge image I3 [ i,j] was obtained according to the Canny operator and the breakpoints in I2 [ i ,j] were connected by comparing with I3 [ i,j]. Finally, the contour was extracted by the contour extraction method to locate the target and the robot was controlled to pick the goods. The experiments show that the method is effective and reliable.

关 键 词：计算机视觉 边缘检测 轮廓提取 视觉机器人 货物拣选 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]