基于图像处理的球笼防尘罩缺陷检测方法

Defect detection method of automobile constant velocity joint boots based on image processing

针对球笼防尘罩的小口内部毛刺缺陷人工质检效率低的问题,提出了一种基于图像处理的实时缺陷检测方法来使质检过程自动化,从而提高生产效率。首先,通过一系列无参数化的图像预处理方法使原始图像满足边缘提取的条件。接着,使用启发式的边缘搜索算法得到小口内部边缘坐标。之后通过最小二乘算法拟合出椭圆参数方程,从而计算得到相似点坐标。最后通过图像边缘与拟合椭圆的相似度来判定物料是否存在小口内部毛刺缺陷。最终实验结果表明,该算法在实际采集的测试数据集中的检测准确率为100%,并且完全可以实现对小口内部毛刺缺陷的实时检测。

A real-time defect detection method based on image processing is proposed to address the low efficiency of manual inspection for intemal burr defects in the small opening of automobile constant velocity joint boots,thereby automating the quality inspection process and improving production fficiency.First,a series of non parametrie image preprocessing methods are applied to ensure that the original image meets the conditions for edge extraction.Second,a heuristie edge search algorithm is used to obtain the intermal edge coordinates of the small opening.Third,the least squares algorithm is employed to fit an ellipse parametric equation,which is then used to calculate the coordinates of matching points.Finally,the presence of intemal bur defects in the material is determined by comparing the similarity between the image edges and the fited ellipse.The experimental results show that the algorithm achieves a 100%detection accuracy on the actual test dataset,and it fully enables real-time detection of internal burr defects in the small opening.