摘要
针对目前无损检测主要采用人工方式存在的主观不一致、检测效率低、操作复杂等问题,设计了一套焊缝缺陷自动检测系统。提出基于Otsu双阈值分割的缺陷区域自动提取、图像的降噪和灰度增强的图像预处理方法;通过SUSAN算法检测焊缝缺陷目标,并结合形态学孔洞填充算法修正缺陷目标;计算焊缝缺陷目标特征参数,并结合所设计的深度为4的二叉树分类识别逻辑流程,实现了较好的焊缝缺陷的检测结果。
Aiming at the problems existing in the nondestructive testing using artificial methods,such as subjective inconsistency,low efficiency and complicated operation,a set of automatic inspection system for weld defects was designed. An image preprocessing method based on Otsu double threshold segmentation for automatic extraction of defective area,image noise reduction and grayscale enhan- cement was proposed ; by SUSAN algorithm for detection of weld defect target, and combined with morphology hole filling algorithm to modify defect targets;through the ealculation of the target characteristic parameters of the weld defects,and combined with the design depth of two of the 4- binary tree classification recognition logic process,to achieve a better detection results of the weld defects.
出处
《电焊机》
2017年第4期89-93,共5页
Electric Welding Machine
关键词
焊缝缺陷
图像处理
特征参数
weld defect
image processing
characteristic parameters
