基于机器视觉的块状工件方位快速识别

Fast recognition of the blocky workpiece position based on machine vision

为了实时准确识别加工生产线上工件的放置方位,提出了一种基于激光三维视觉检测的块状工件方位快速识别方法。对采集的工件激光图像,采用快速均值滤波算法进行预处理以抑制散斑噪声,并给出了交叉激光迹线的分离和快速细化处理方法。为了提高图像匹配的速度和精度,在模板和待测曲线配准上,定义了两种距离测度作为配准的评价函数。试验结果表明,该方法识别正确率高,速度快,对工件的角度偏移和平移不敏感;较好地满足了生产线上的实时检测要求。

In order to quickly and accurately identify the blocky workpiece position in the production line,a method based on laser 3D visual inspection is proposed. For the captured images,a fast median filtering algorithm is used to suppress the speckle noise and then a fast separation and refinement algorithm of the crossover laser traces is presented. On the template matching process,two kinds of distance measure to improve the image registration rate are defined. The experimental resulsts show that the method is not only of higher recognition rate but also of faster speed,at the same time,which is not sensitive to the angular displacement and the translation,perfectly meeting the requirements of real-time detection of the production line.