Abstract: According to the welding characteristic of container reinforced plate，an automation positioning method based on the structured light of the welding seam is developed.Two traits are included in the method：The camera is located on a right-angled framework and its optical axis is normal to the working plane，then depth information can be obtained by comparing the size of area，accordingly three-dimensional information extraction can be transformed into two-dimensional and space point can be transformed image point by affine transform.This system consists of two lasers and one camera.It will be much easier to process image because of the significant characteristic of laser light in welding environment when laser is used as an aide to help camera to locate much reliably.The particular implementation procedure is as follows：Transition emerged on the welding between the reinforced plate and the roof plate when two laser beans are projected to an angel of the reinforced plate.Four couple of image coordinates of transition points are attained by processing tripe effectively which are used to gain the reinforced plate’s welding initial characteristic point and orientation by fitting a straight line，finally the other characteristic points can be attained one by one referring to the image template which is made beforehand.This system takes advantage of the laser efficiently，resolves the automatic positioning problems on reinforced plate welding beam reliably.

Key words: vision positioning, structured light, affine transform, image processing

摘要： 针对集装箱顶角加强板的焊接特点，设计了一种基于结构光的自动焊缝定位方法。该方法主要有两个特点：一是将摄像机安装在机器人上，使光轴垂直工作平面，通过比较面积大小来获得深度信息，从而将三维信息的获取问题转化为二维信息的提取，并利用仿射变换获得空间信息;二是利用激光器与摄像机相配合的方式进行定位，由于激光束在焊接环境中的特征明显，可降低图像处理的难度。具体实现过程是利用两条线激光打在加强板的一个角上，在加强板和顶板之间的焊缝处出现转折，通过对激光条纹的有效处理，获得4对转折点的图像坐标，然后拟合直线求出加强板的焊接初始位置及方向，再结合事先制作的图像模板将加强板的其他特征点逐一求出。该系统有效地利用了激光器的优势，可靠地解决了加强板焊缝的自动定位问题。

关键词: 视觉定位, 结构光, 仿射变换, 图像处理