摘要
为了解决当前焊缝缺陷检查算法在焊缝缺陷特征微弱、背景复杂的环境下,导致其缺陷漏检的不足,本文设计了基于水平集与强分类器Adaboost的焊缝图像缺陷检查算法。首先,设计水平集迭代轮廓线,制定演化收敛停止条件,定位出焊缝准确位置,并提取其轮廓;然后计算轮廓几何特征,建立基于Ada Boost的分类器,完成对缺陷的确认和分类。实验数据显示:与当前工件表面焊接缺陷检查技术相比,面对缺陷特征微弱、背景复杂环境时,本文算法具有更高的检测精度,能准确定位识别焊缝缺陷。
In order to solve the current weld defect inspection methods are weak in weld defect under the background of complex environment leads to the defect leak, this paper designes the Adaboost classifier based on level set to steel weld defect detection algorithm. First of all, the level set iterative contour design, makes evolutionary convergence conditions precedent, locates an outline of the weld position accurately extracts. The geometric characteristics are calculated, based on Adaboost classifier, recognition and classification of defects. Experimental data shows that： compared with the current steel weld defect inspection technology, in the face of weak defect features, complex background environment, this algorithm has higher precision of inspection.
出处
《自动化技术与应用》
2016年第10期95-99,共5页
Techniques of Automation and Applications
关键词
焊缝缺陷检测
水平集
强分类器
几何特征
演化收敛
weld defect detection
level set
Adaboost
geometric feature
evolution of convergence
