基于对数极坐标变换的X射线焊缝缺陷图像识别算法  

Algorithm for Detecting X-ray Welding Defects Image based on LPT

作　　者：丛军 张辉宇 谢建 陈健飞 高海涛 张宏 韩曙光 王征 CONG Jun;ZHANG Huiyu;XIE Jian;CHEN Jianfei;GAO Haitao;ZHANG Hong;HAN Shuguang;WANG Zheng(Shengli Offshore Petroleum Engineering Technology Inspection Co.,Ltd.,Sinopec,Dongying 257000,Shandong,China)

机构地区：[1]中石化胜利海上石油工程技术检验有限公司,山东东营257000

出　　处：《焊管》2022年第2期39-43,共5页Welded Pipe and Tube

基　　金：陕西省重点研发计划项目“冷阴极X射线焊缝缺陷自动检测系统”(项目编号2020GY-179)。

摘　　要：为了提高X射线焊缝图像缺陷分类识别的成功率,提出利用对数极坐标变换(LPT)的距离及角度不变特性解决实际缺陷识别中因缺陷图像位置、形状及方向变换导致的识别准确率下降问题。同时,为降低噪声的影响,提出基于压缩传感技术的缺陷识别,将待检测图像视为一组样本图像的线性组合,通过求取系数向量确定待检测图像的类型。由于缺陷的信息存于一组系数向量中,而非依赖于某几个特定的特征值,因此任一图像的变化对最终识别结果不会造成影响。实际试验表明,LPT结合压缩传感技术可以实现较高的X射线焊缝图像缺陷识别率,极大地降低了噪声误报率。In order to improve the success rate of X-ray welding defect identification,the distance and angle invariance of log polar coordinate transformation(LPT)is proposed to solve the problem of accuracy degradation caused by the transformation of defect image position,shape and direction in practical defect recognition.At the same time,in order to reduce the influence of noise,a defect recognition based on compressed sensor technology is proposed.The image to be detected is regarded as a linear combination of a group of sample images,and the type of image to be detected is determined by calculating the coefficient vector.Because the information of defect image is reflected by a group of coefficient and not by some specific characteristics,any change in one image has no influence on the recognition result.Experiments show that LPT combined with compress sensor can achieve high defect recognition rate of X-ray weld image and greatly reduce the noise false alarm rate.

关 键 词：缺陷识别 X射线 压缩感知 对数极坐标变换 

分 类 号：TG441.7[金属学及工艺—焊接]