摘要
为解决电池模组极柱焊接缺陷检测精度低、效率低的问题,提出了一种基于机器视觉的焊接缺陷检测算法。首先,对采集图像进行预处理操作;其次,通过组件筛选结合改进的Canny算法获取目标区域的无干扰边缘轮廓,为了改善拟合干扰现象,利用基于密度的聚类(density-based spatial clustering of applications with noise,DBSCAN)算法对焊接区域的内外圆边缘点集进行分离;然后,采用改进的最小二乘法对内外圆点集分别进行拟合得到精准的焊接区域;最后,以焊接区域内外圆的面积差和同心度来检测焊接面积缺陷和焊偏,通过双向扫面检测法进行焊接区域灰度值遍历,根据对应的灰度值范围和区域大小来检测焊穿和炸点缺陷。实验表明,该算法能够精确拟合焊接区域并准确识别出焊接缺陷,具有较高的检测精度和效率,能够满足工业生产需求。
To solve the problem of low precision and efficiency in battery module terminal welding defect detection,a welding defect detection algorithm based on machine vision was proposed.Firstly,the acquired image is preprocessed,then the target area′s interference-free edge contours were obtained through component selection combined with an improved Canny algorithm.To improve the fitting of interference,the DBSCAN(density-based spatial clustering of applications with noise)algorithm was used to separate the inner and outer circular edge point sets of the welding area,and then the inner and outer circular point sets were fitted using an improved least squares method to obtain precise welding areas.Finally,the welding area defects of area and weld offset were detected by the difference in the area of the inner and outer circular regions and the concentricity,and the welding defects of weld-through and explosion were detected by the bidirectional sweep detection method through gray value traversal of the welding area.According to the corresponding gray value range and area size,the defects were detected.The experimental results show that the algorithm can accurately identify welding defects and precisely fit the welding area,with high detection accuracy and efficiency,which can meet the needs of industrial production.
作者
陈甦欣
姚俊杰
赵毅
CHEN Suxin;YAO Junjie;ZHAO Yi(School of Mechanical Engineering,Hefei University of Technology,Hefei 230009,China)
出处
《组合机床与自动化加工技术》
北大核心
2025年第5期136-139,144,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
安徽省重点研究和开发计划项目(202304a05020079)。
关键词
机器视觉
聚类
最小二乘法
圆拟合
缺陷检测
machine vision
cluster
least square method
circular fit
defect detection
