基于图像纹理与矩特征的转子绕线检测研究

摘要

传统电机转子生产过程中,转子挂钩处绕线的合格情况主要依靠人工肉眼检测,受疲劳、环境等因素影响,检测效率较低且易出错.以电机转子图像为样本进行自动检测算法的研究,对采样的工件图像进行定位、分割、滤波等预处理,再与标准模板进行相似度匹配,判断合格性.针对标准匹配模板选取的偶然性问题,提出先基于图像纹理特性进行定量描述,再根据欧式距离对合格模板进行筛选,最后基于图像矩特征进行相似度匹配的方法.实验结果表明,这一方案既提高了运算效率又保证了识别率.%The qualification of rotor winding from the hook of motor mainly relies on the artificial visual inspection during the process of motor production in current factory.The detection efficiency was low and error-prone due to the factors of fatigue and environment.The automatic detection algorithm of motor rotor image was studied.The images of the workpiece were preprocessed by positioning,segmentation,filtering according to the digital image processing technology,and matched with the standard template similarity to determine the eligibility.To solve the contingency problem of standard matching templates selection,the standard images were described quantitatively based on the texture characteristic and then screened according to the Euclidean distance.Finally,the similarity matching were conducted based on the moment features.The experimental results show that the proposed scheme improves the computational efficiency and ensures the recognition rate.