倒角类冲压零件轮廓缺陷视觉检测

Vision Detection of Small Contour Defects of Chamfering fine-blanking parts

摘要:

倒角类冲压零件轮廓缺陷视觉检测对于提高生产效率、保障产品质量非常重要。针对倒角类冲压零件形状奇异、轮廓不规则,图像本身实物边缘与背景像素是逐渐过渡的,存在模糊性,缺陷特征弱小等难题,提出了模糊集合的轮廓提取算法和缺陷匹配的识别方法,检测并标记缺陷位置。首先,采用模糊集合强化边缘信息,在不需确定阈值的情况下,准确有效地提取冲压零件的轮廓。其次,利用轮廓的点、线特征,通过HOUGH变换对冲压零件进行定位。然后,以冲压件模板图像位姿特征为期望值,实时校正待检测的冲压件图像,实现精确对准。最后,采用差分相似匹配算法识别出缺陷。实验结果表明,提出的检测方法能够快速、准确地识别占整个零件0.4%以上的小缺陷,并标记位置,满足工业要求。

Abstract:

Contour defects detection of the chamfering fine-blanking parts is very important to increase productivity and ensure products desired quality. However, the contour of chamfering fine-blanking parts are irregular, the image itself exists ambiguity due to pixels of target’s edge and background with gradual transition, and the feature of contour defect is very weak. Thus the contour extraction algorithm and a defect recognition method with fuzzy sets are proposed to detect defects and mark the location. Firstly, the fuzzy set is employed to strengthen the edge information, suppress noises and extract the contour accurately and effectively. According to the set of points and line features of the extracted contour, the method of obtaining location and attitude of stamping parts through HOHGH transformation are designed. And then, the location and attitude of stamping parts to be detected will be aligned precisely in real time using the pose feature of the stamping template image. The difference similarity matching algorithm is utilized to identify the defects. Some experimental results show that more than 0.4% small defects of the total part can be identified and located quickly and exactly which meets industrial inspection requirements.