基于视觉显著性的零件缺陷检测

Defect Detection of Mechanical Parts Based on Visual Saliency

为了提高零件缺陷检测的准确率,提出了一种基于视觉显著性算法的零件缺陷检测方法。首先将采集零件缺陷图像进行高斯差分滤波,以最大程度消除背景信息的干扰。然后对高斯差分滤波后的零件缺陷图像进行超像素分割,并利用全局图像对比方法构建超像素图像显著图,从而有效的提高缺陷的显著性。最后,采用最大类间方差法分割缺陷。试验表明该方法能提高零件缺陷的检测准确率。

In order to improve the accuracy of part defect detection,a novel method of part defect detection based on visual saliency algorithm is presented.Firstly,the defect image was filtered by Gauss difference to eliminate the interference of background information.Then,the part defect image after Gauss difference filtering was segmented by super-pixel,and the super-pixel saliency image was constructed by using the global image contrast method,which effectively improves the defect saliency.Finally,the method of maximum inter class variance was used to detect defection.The experiment results show that the method can improve the detection accuracy of part defects.