一种V型坡口焊缝图像处理方法及特征提取算法研究

A Pre-Processing and Feature Point Extraction Algorithm for V-Groove Weld Images

为了提高V型坡口焊缝特征提取算法的效率和准确性,对V型坡口光条图像预处理,在图像感兴趣区域(Region of Interest,ROI)开窗基础上构建差异化卷积模板,经图像差分后实现光条纹增强及噪声颗粒化.通过形态学开运算和小连通域去除提取出二值化光条,并采用几何中心法完成光条骨架细化.通过对光条形态学特征分析,初步定位角点,分区域提取光条中心线,最终获得焊缝的精确特征点.实验结果表明,采用该方法能够有效去除噪声,准确提取出亚像素级特征点,相较传统角点检测算法效率提升60%,满足工业应用的高精度和实时性要求.

The V-groove stripe images is preprocessed to improve the efficiency and accuracy of the algorithm for feature extraction of V-groove welds.The differential convolution templates are constructed on the regions of interest(ROI),the laser stripe enhancement and noise granulation are realized after image difference.The binarized laser stripe is extracted by morphological opening operation and small connected domain removal,and the stripe skeleton refinement is completed by the geometric center method.The morphological characteristics of the V-groove stripe is analyzed,and the stripe center-lines are extracted in different regions rely on calculation of corner points.The precise feature points of the weld are obtained.The experimental results show that this method can effectively remove noise and accurately extract sub-pixel feature points.The efficiency of the proposed algorithm is 60%higher than traditional corner detection algorithms,which meets the high-accuracy and real-time requirements of industrial applications.