基于机器视觉的轴承识别与定位算法研究

Research on Bearing Recognition and Positioning Algorithm Based on Machine Vision

针对运用视觉技术对轴承进行质量检测与尺寸测量时,能够准确地识别定位到目标轴承,提出一种基于机器视觉的轴承识别与定位算法。通过对采集到的轴承图像进行预处理,分割出目标图像并提取图像的外轮廓边缘特征;设置轴承的模板图像,结合图像的Hu不变矩特征对轴承进行识别匹配;通过最小二乘法对图像边缘点进行圆拟合并采用迭代法进行修正,通过计算圆心的位置坐标,实现对轴承的定位。实验结果为:轴承的识别匹配度在0~0.03之间,定位误差在0.5像素以内,满足系统对轴承的识别定位精度要求。

Given that the vision technology can accurately identify and position target bearing in its application of bearing quality inspection and size measurement,a bearing identification and location algorithm based on machine vision was proposed.By preprocessing the collected bearing image,the target image was segmented and the outer contour edge features of the image were extracted.The template image of the bearing was set,and by combining with Hu moment invariant features of the image,the bearing was recognized and matched.The circle fitting of the image edge points was set by the least square method,and the iterative method was adopted for correction.By calculating the position coordinates of the center of the circle,the positioning of the bearing was realized.The experimental results show that the recognition matching degree of vehicle bearing is between 0 and 0.03,and the positioning error is within 0.5 pixel,which meets the requirements of the system for the accuracy of bearing recognition and positioning.