精冲件轮廓缺陷在线检测技术

Online detection technology on contour defects for fine blanking parts

为提高平板厚板型精冲件的轮廓检测速度与精度,基于机器视觉技术,研究了一种轮廓缺陷在线检测技术。利用Canny算子从均值偏移滤波后图像中提取边缘线,然后根据标准轮廓(模板)的尺寸,从中挑选出需要检测的轮廓。由于待检测轮廓与模板存在角度与位移偏差等问题,提出了一种先角度配准、再位置配准的两步图像配准算法,将配准后的待检轮廓与模板进行图形比对,差异图像经过形态学滤波等算法处理,获得缺陷区域的尺寸及位置,由此实现精冲件的在线检测。基于研究成果设计开发了一套零件在线检测系统,并进行了实验验证。结果表明,该系统能够识别的零件精度达0.4 mm,每件的识别时间小于0.3 s,完全能够满足大批量精冲零件的轮廓缺陷在线检测的需求。

In order to improve the speed and accuracy of contour detection for fine blanking parts of thick and flat plates,an online detection technology for contour defects was studied based on machine vision technology.Then,the edge lines were extracted from the mean shift filtered image by the Canny operator,and according to the size of standard contours(templates),the contours to be detected were selected.Due to the problems of angle and displacement deviation between the contour to be detected and the template,a two-step image registration algorithm was proposed,which first registered the angle and then registered the position,and the registered contour to be detected was graphically compared with the template.Furthermore,the difference image was processed by the algorithms such as morphological filtering to obtain the size and location of the defect area,thereby realizing the online detection of fine blanking parts.Finally,based on the research results,a set of online detection system for parts was designed and developed,and the experimental verification was carried out.The results show that the system can recognize parts with the accuracy of 0.4 mm,and the recognition time of each part is less than 0.3 s,which can fully meet the needs of online detection for contour defects in mass fine blanking parts.