基于Canny算子和形态学滤波的焊缝图像背景去除技术

Background removal technology of weld image based on Canny operator and mathematical morphology

为了减少焊缝图像中焊道以外的冗余数据对缺陷检测造成影响,提出一种基于Canny算子和形态学滤波的焊缝X射线图像背景去除技术。原始焊缝图像使用中值滤波平滑后先通过Canny算子得到图像的粗略边缘,再使用数学形态学中的膨胀和细化使焊道边缘闭合。最后通过区域生长算法生成焊道区域灰度值为1,背景灰度值为0的二值化掩膜图像,该图像与原图像相与得到背景去除后的焊缝图像。实验结果表明,该算法处理后图像的轮廓纯净度和完整性都得到了较大的改进,前景边缘更加清晰,背景去除更加彻底,有效地避免了背景对焊缝检测的影响。

In order to reduce the influence of redundant data on weld defect detection,a background removal technique for weld X-ray image based on Canny operator and mathematical morphology is proposed.First of all,the original weld image is smoothed by median filter and then the rough edge of the image was obtained by Canny operator.Secondly,the edge of weld bead is connected by expansion and refinement algorithm of mathematical morphology.Finally,a binary mask image with weld bead area whose gray value is 1 and background area whose gray value is 0 is generated by region growth algorithm.The weld image without background is obtained by the original image multiplied by the mask image.Experimental results show,the performance of the proposed algorithm is greatly improved in terms of outline purity and integrity.It means that the foreground edge is clearer,the background removal is more thorough,and the influence of background on weld detection is effectively avoided.