基于机器视觉的机械零件表面缺陷在线检测方法研究

Research on On-line Detection Method of Mechanical Parts Surface Defects Based on Machine Vision

利用传统机器视觉来检测零部件的表面缺陷已经无法满足零部件标准化检测,尤其是零部件光滑表面形成的强烈反射效应,造成零部件的部分表面亮度饱和,检测的图像信息缺失,精度严重下降。针对检测精度下降的问题,提出一种基于差影模型重构的方法,引入缺陷区域边界近似系数和细节系数来提升整个零部件缺陷检测精度。为了验证所设计的表面缺陷检测方法的准确性与可靠性,设计了在线检测试验平台,结果表明该方法能够高效地提高零部件表面缺陷检测的精度。

The use of traditional machine vision to detect the surface defects of components can’t meet the standardized detection of components,especially the strong reflection effect formed by the smooth surface of the components,resulting in the saturation of part of the surface of the components,the lack of detected image information,and the serious decline in accuracy.Aiming at the problem of reduced detection accuracy,a method based on the reconstruction of the differential shadow model is proposed,and the boundary area approximation coefficient and detail coefficient of the defect area are introduced to improve the detection accuracy of the entire part defect.In order to verify the accuracy and reliability of the designed surface defect detection method,an online test platform is designed.The results show that the method can effectively improve the accuracy of surface defect detection of parts.