多线激光V型焊缝图像特征提取与轨迹识别

为了减少焊缝识别中传感器安装位置带来的前置误差,提高识别精度,设计了一款多线激光传感器,对多线激光在V型焊缝上的图像进行处理得到焊缝轨迹。提出一种基于动态不规则感兴趣区域的图像分割方法,将多条激光分割后并行处理。单条激光线图像预处理后以像素行灰度值之和的极大值确定激光条纹所在行,对激光条纹分段拟合后求交点,得到坡口边缘与焊缝位置的特征点,并将焊缝位置特征点拟合得到焊缝轨迹。根据所提出的算法编写了焊缝轨迹识别程序,并进行实验,结果表明,所提出的算法识别焊缝走向误差在0.5°之内,符合焊缝跟踪的要求。

Feature Extraction and Trajectory Recognition of V-Groove Weld Seam Based on Multiline Laser

In order to decrease the error that caused by the position of sensor,improve the precision of weld seam recognition,a multiline laser sensor is designed to process the multiline laser image on the V-groove weld sead to acquire the trajectory of weld seam.An image segmentation method based on irregular dynamic region of interest is proposed to divide lasers and process image parallelly.After preprocessed the image of single laser,the row of laser was located by the extremum of gray value sum.After fitting the laser liner in sections,the points of intersection were obtained,and the feature points of groove edge and weld seam position were obtained,and the trajectory of weld seam was obtained by fitting the feature points.Based on the proposed algorithm,recognition program of weld trajectory was compiled,then an experiment was conducted.The results indicate that the proposed algorithm identify the weld direction error is whitin 0.5° and meets the requirement of weld tracking.