工字形焊件射线图像缺陷分割中残余噪声的去除方法
Methods for Removing Residual Noises in Defects Segmentation for X-ray Detection Images of I-shaped Weldments
石端虎1， 刚 铁2， 李万里1， 杜成超1
点击：3251次 下载：0次
作者单位：（1. 徐州工程学院 机电工程学院，江苏 徐州 221008；2. 哈尔滨工业大学 现代焊接生产技术国家重点实验室，黑龙江 哈尔滨 150001）
中文关键字：工字形焊件； 射线图像； 缺陷分割； 残余噪声; 去除方法
英文关键字： I-shaped weldments； X-ray image； defects segmentation； residual noises； remove methods
中文摘要：采用左右旋转焊件的方法获得工字形焊件的射线检测图像，对其缺陷与背景间的灰度分布特征进行了统计分析，结果表明，要实现图像中小缺陷的分割有很大的难度。本文提出了图像中小缺陷分割的算法流程，研究了图像分割过程中残余及孤立点噪声的去除方法；并采用剖切焊件的方法对分割结果进行了试验验证，试验结果表明提出的方法是可行的。
英文摘要：The X-ray detection images of I-shaped weldments were obtained by left-right rotation welding. Statistics analysis of gray distribution features between defects and its background was proceeded. The experimental results show that it is very difficult to achieve the segmentation of small defects. The segmentation algortithm procedure for small defects was put forward, the method for removing residual and single noises in the course of image segmentation was studied. The destructive experiments of I-shaped weldments were carried out to verify the segmentation results. The results show that the provided method is feasible.