基于去辊道背景实现工件位姿识别

A Method of Identifying the Workpiece Pose Based on Roller Table De-Background

针对工业辊道高干扰环境中识别无序摆放工件时存在的识别准确率低与耗时长等问题,提出一种新型去背景提取工件最小包围轮廓方法(RBEC)去获取辊道上无序摆放工件的角度和位置信息。首先,用滤波后的背景图像与工件图像进行像素相减去除阴影与大部分干扰背景;其次,通过提出的辊道残留去除算法除去图像背景中残留的滚筒边缘;最后,采用形态学闭运算连通只剩工件区域的像素点并通过提取该区域的最小包围轮廓信息获取辊道上的工件角度和位置信息。在实际工业环境中的工件分拣平台上进行实验,其结果可满足位姿识别需求,且RBEC法计算简单,鲁棒性高,相较于SIFT法和SURF法,RBEC法计算时间分别降低88.7%和83.0%,角度误差分别下降了45.9%和44.5%。

To solve the problems of low recognition accuracy and long time-consuming in identifying disordered workpieces in industrial roller table high interference environment,a new background removal and minimum contour extraction method(RBEC)is proposed to obtain the angle and position information of disordered workpieces on the roller table.Firstly,the roller table image with the workpiece is blurred by the maximum and minimum value,and the shadow on the edge of the workpiece is removed by pixel subtraction with the original image.At the same time,after the above processing of the empty roller image,the empty roller image and the filtered workpiece image are subtracted to remove most of the interference background.Secondly,the roller edge in the image background is removed by the proposed roller edge removal algorithm.Finally,the morphological closed operation is used to connect the pixels of the remaining workpiece area,and the angle and position information of the workpiece on the roller table is obtained by extracting the minimum contour information of the region.The experiment is carried out on the workpiece sorting platform in the actual industrial environment,and the results can meet the needs of pose recognition,and compared with sift method and SURF method,the calculation time of RBEC method is reduced by 88.7%and 83.0%respectively,and the angle error is reduced by 45.9%and 44.5%respectively.