【摘要】 引入Gabor小波方法去除光照干扰,实现汽车金属材料表面微小缺陷的快速识别。通过改进Otsu阈值分割法对缺陷图像进行分割,对缺陷图像实行进一步处理去除光照干扰;利用模拟测试法得到Gabor滤波器运行中的最佳参数,对样本汽车金属材料表面微小缺陷图像进行Gabor模板卷积操作,获取边缘图像;针对边缘图像实行加权马氏距离计算,对边缘轮廓特征进行增强;通过连通区域标记检索并标记汽车金属材料表面的缺陷,实现汽车金属材料表面微小缺陷的快速识别。研究结果表明:Gabor小波的缺陷识别方法缺陷识别率高,能够准确识别横向划伤、纵向划伤以及网纹等缺陷,具有较高的可行性。更多还原

【Abstract】 Gabor wavelet method was introduced to remove the illumination interference and realize the rapid recognition of micro defects on the surface of automotive metal materials. The improved Otsu threshold segmentation method was used to segment the defect image, and the defect image was further processed to remove the light interference. The simulation test method was used to get the best parameters in the operation of Gabor filter, and the Gabor template convolution operation was performed on the sample automobile metal material surface micro defect image to obtain the edge image. The weighted Mahalanobis distance was calculated for the edge image, and the edge contour feature was enhanced. The defects on the surface of automotive metal materials were retrieved and marked by the connected region marking, so as to realize the rapid recognition of micro defects on the surface of automotive metal materials.The results show that the Gabor wavelet defect recognition method has high recognition rate, and can accurately identify the defects such as horizontal scratch, vertical scratch and mesh, so it has high feasibility.更多还原