基于机器视觉转子冲片亚像素精度尺寸测量研究

Research on sub-pixel precision dimension measurement of rotor punching based on machine vision

摘要:

针对传统冲压件人工尺寸测量效率低、精度难以保证等问题，提出一种基于机器视觉的转子冲片亚像素精度尺寸测量方法。该方法采用先将采集的冲片图像灰度化再应用中值滤波来降低噪声干扰的图像预处理方法；通过Canny算子与Otsu方法相结合实现自适应阈值边缘检测，再利用改进的Zernike矩方法进行亚像素定位，获取亚像素级坐标；然后设计轮廓分割算法，主要提取内外圆亚像素轮廓，同时设置感兴趣区域并基于K-Means聚类算法分割出骨架线段轮廓，最后使用最小二乘拟合法求解出转子冲片内外圆直径和骨架间距尺寸。实验结果表明，该方法平均测量精度可以达到0.01mm，测量精度高、速度较快，具有较高的实用价值。

Abstract:

Aiming at the problems of low efficiency and difficult to guarantee the accuracy of traditional stamping parts manual dimension measurement, a sub-pixel precision dimension measurement method for rotor punching based on machine vision is proposed. This method adopts the image preprocessing method of first graying the collected image and then applying the median filter to reduce the noise interference. The adaptive threshold edge detection is realized through the combination of Canny operator and Otsu method, and then the improved Zernike moment method is used for sub-pixel positioning to obtain sub-pixel coordinates. Then design the contour segmentation algorithm, which mainly extracts the sub-pixel contour of the inner and outer circles. At the same time, set the region of interest and segment the outline of the skeleton line segment based on the K-Means clustering algorithm. Finally, use the least squares fitting method to solve the diameter of the inner and outer circles of the rotor punch and the skeleton Spacing size. The experimental results show that the measurement accuracy of this method can reach 0.01mm, the measurement accuracy is high, the speed is fast, and it has high practical value.