摘要
针对表面检测中存在的问题,文中以研磨工艺的电连接器壳体为研究对象,提出了一种基于机器视觉的表面检测方法。使用基于小波分析的方法对表面缺陷进行检测,继而对无表面缺陷的壳体检测表面粗糙度检测。提取不同等级表面粗糙度比较样块图像的8个灰度共生矩阵特征,进行最小二乘法直线拟合,依据拟合优度,择优建立表面粗糙度关系模型,并根据该模型对待测壳体表面粗糙度进行实验分析。实验结果表明,该模型的表面粗糙度检测误差在7%以下。
Aiming at the problem of surface detection, this paper proposes a set of surface detection method based on machine vision. The surface defects are detected by using wavelet analysis method, and then the surface roughness of the shell without surface defects is detected. 8 GLCM features are extracted from different levels of sur- face roughness of comparative sample image, linear fitting with least square method, basis of goodness of fit, pre- ferred to establish the relationship model of surface roughness, and according to the model treats the measured shell surface rough degree of experimental analysis. The experimental results show that the model of surface roughness detection error is below 7 %.
出处
《电子科技》
2017年第3期146-148,152,共4页
Electronic Science and Technology
关键词
机器视觉
壳体
表面质量
小波
灰度共生矩阵
machine vision
electric connector shell
surface quality
wavelet
GLCM
