基于尺度乘积的X射线焊缝区域提取算法研究  

Weld Region Extraction in Radiographic Image Based on Scale Multiplication Technique

作　　者：匡平[1] 张明星[1] 万维[1] 

机构地区：[1]电子科技大学计算机科学与工程学院成都611731

出　　处：《电子科技大学学报》2015年第5期737-742,共6页Journal of University of Electronic Science and Technology of China

基　　金：四川省科技支撑计划(2013GZX0155,2013GZX0165)

摘　　要：焊缝缺陷自动检测是保证焊接质量的重要环节,而焊缝区域的提取是焊缝缺陷特征提取与焊缝缺陷识别的基础。该文提出了一种基于尺度乘积的X射线焊缝区域提取算法。该方法首先对图像按行取样获得行灰度曲线,再利用最小二乘直线拟合的方法将不同尺度下拟合直线的斜率乘积代替梯度算子,并进行非极大值抑制得到边缘点。对图像每行进行同样的操作,得到整个焊缝的边缘,从而提取出焊缝区域。实验表明,该方法能显著抑制噪声的干扰、提高边缘检测的准确率,在噪声多、边缘模糊的X射线图像中能取得比主动轮廓模型和模糊核聚类更好的效果。Automatic detection of weld defects is an important part of the welding quality assurance, while the extraction of weld region is the base of feature extraction and defects recognition. In order to accurately extract the weld region, this paper presents an X-ray weld region extraction algorithm based on scale multiplication. In this method, the grayscale plot is obtained by line scan of the image, and then instead of a gradient operator, the product of slopes of fitting lines in different scales, which are computed by the least squares linear fit, are used for non-maxima suppression to get edge points. Applying the same operation to each line, we can obtain the whole picture of the weld edges to extract the weld region. Experimental results show that this method, in X-ray images with noise and blurred edge, can significantly suppress noise interference, improve the accuracy of edge detection, and can acquire better effect than the implicit active contours and the kernel fuzzy c-means.

关 键 词：边缘检测 灰度曲线 最小二乘直线拟合 尺度乘积 焊缝区域提取 

分 类 号：TP393.02[自动化与计算机技术—计算机应用技术]