基于B样条曲线的X射线图像焊缝缺陷分割与提取

Weld defect segmentation and extraction of X-ray image based on B-spline curve

针对大部分X射线数字图像低对比度、背景起伏大以及纹理复杂的问题,在X射线数字焊缝图像预处理之后,采用B样条曲线对列灰度曲线进行拟合,获得光滑而且顺畅的曲线.在此基础上进一步提取曲线的极值点,并通过定义的波动阈值以及边界阈值进行两次极值点集合的修正.最后利用数学形态学及中值滤波对缺陷的形状和大小进行了修正.结果表明,该技术有效地解决了X射线图像由于焊缝纹理复杂导致缺陷提取困难的问题,有利于实现X射线图像焊接缺陷的自动提取.

Aiming at the facts that most X-ray digital images have the problem of low contrast,large background fluctuation and complex texture,b-spline curve is used to fit the column gray curves after preprocessing of weld images.The result shows that redundant curve peaks are eliminated and a smooth curve is abtained.Furthermore,curve extreme points are extracted and extreme value point set is modified twice by a fluctuation threshold and a boundary threshold.Finally,the shape and size of defects are revised by mathematical morphology and median filtering.Experiment results shows that this strategy effectively solves the problem of defect extraction caused by the weld seam with complex texture in X-ray image.It is advantageous to realize the automatic extraction of weld defects in X-ray image.