冲压磨削平板类零件的表面缺陷检测

Surface defect detection on stamping and grinding flat parts

为实现机器视觉在冲压磨削平板类零件表面缺陷检测的应用,开发了一套在线检测系统。磨削加工后,零件表面有走向一致、深浅不一的磨削刀纹。实验发现,平面光照下零件整体区域清晰、轮廓明显,但细微缺陷难以发现;有向光可以突出细微缺陷,但刀纹反光,造成图像质量较差。针对以上情况,设计了一种平面光与有向光结合的打光方式,以利用他们各自的优点。根据缺陷分布特征将零件区域分为内、外两部分分别寻找缺陷,采用模板匹配、阈值分割、特征提取等方式进行图像处理。针对刀纹反光影响,提出了一种基于边缘寻找缺陷的方法,根据灰度异常区域与边缘线的位置、方向判断该异常是由缺陷引起还是由反光引起。经实验验证,本方法检测的准确率可达97%。

In order to realize the application of machine vision in surface defect detection of stamping and grinding flat parts,the on-line inspection system was developed.After the grinding process,the grinding blade patterns on the surface of parts had the same direction and different depth.It was found that the clear whole area and the obvious outline of part under plane illumination were observed by experiment,but the subtle defects were difficult to find.Then,the directed light highlighted the subtle defects,but the blade pattern was reflective resulting in poor image quality.For the above situations,the lighting method combining plane light and directed light was designed to take advantage of their respective advantages.According to the characteristics of defect distribution,the part area was divided into internal and external parts to find the defects,and the image was processed by template matching,threshold segmentation and feature extraction.Furthermore,for the influence of blade pattern reflection,the method of finding defects based on edges was proposed,and it was determined whether the abnormality was caused by defect or by light reflection according to the positions and directions for the abnormal grayscale regions and the edges.The experiments show that the accuracy of this method is up to 97%.