基于工业机器人弧焊的焊缝识别研究

Research on Weld Identification Based on Industrial Robot Arc Welding

从视频或图像序列中准确有效提取出焊缝位置信息是实现工业机器人自主弧焊的基础和研究关键。根据主动视觉识别技术,利用CCD摄像机和激光视觉传感器搭建硬件系统,进行焊缝图像采集,提出了一种基于图像预处理与后处理相结合的焊缝检测方法,以适应于复杂焊接环境情况下的焊缝识别。对原始采集图像做图像滤波等预处理工作滤除噪声和提高焊缝图像清晰度,通过改进图像分割和孤点滤波滤等图像后处理提取出焊缝感兴趣区域,最后通过几何分析法和斜率法提取出焊缝中心线和特征点,得到准确的焊缝位置特征信息,从而完成图像信息到焊缝位置信息转换。实验仿真结果表明,所提出算法在搭建硬件系统上的焊缝识别率达到92%,能够把误检率控制在11%左右,能够在复杂背景情景下准确提取出焊缝位置信息,满足工程应用要求。

Accurate and effective extraction of weld position information from video or image sequences is the basis and key to realize the independent arc welding of industrial robots. According to the active visual recognition technology, the CCD camera and laser vision sensor are used to build the hardware system to weld the weld image. A weld seam detection method based on image pre-processing and post-processing is proposed to adapt to the complex welding environment. Weld seam identification. Perform pre-processing work such as image filtering on the original acquired image to filter out noise and improve the sharpness of the weld image. The image is extracted by image post-processing such as image segmentation and orphan filter filtering to extract the region of interest of the weld, and finally pass the geometric analysis method and slope. The method extracts the weld center line and feature points, and obtains accurate weld position feature information, thereby completing image information to weld position information conversion. The experimental results show that the proposed algorithm can achieve a weld seam recognition rate of 92% on the hardware system, and can control the false detection rate to about 11%. It can accurately extract the weld position information under complex background scenarios to meet engineering application. Claim.