基于激光视觉系统的水下焊接焊缝识别

Recognition for underwater welding seam based on laser vision system

设计了一套基于激光器的视觉系统,该系统利用激光器作为辅助光源,同时利用复合滤光片进行滤光,由CCD采集图像。设计了一种基于最大灰度平均值及其改变量的二维输入、一维输出模糊处理器,能自动获取每个3×3列区域的局部阈值,二值化后单独求取焊缝图像中每个光斑的平均高度,提出基于贝叶斯最小概率模式识别对V形焊缝的左右边进行分类,通过准确分类获得V形焊缝左右边的平均高度,以此高度为基准,结合焊缝深度的特征可以在空域上最大限度去除强光斑形成的干扰数据;最后用分类获得的数据对焊缝的左右边采用最小二乘法进行拟合,可以准确获取焊缝中心位置的坐标。

A vision system based on laser was designed in this paper,which used a laser as assistant lamp-house with compound filters to filter, and weht images with CCD is collected for the system.A fuzzy processor that had two dimensions inputs which were maximum average gray value and its change value was designed,the processor had a dimension output which was the loeal threshold c^t＇ the region when a program scanning with 3x3 pixels-window.EvetLy faculae＇ average heights were obtained separately by calculating after making threshoht processing to a weht image,meanwhile,to classify the left and the right sides of the V-groove weld,a kind ot mode recognition based on the minimum error probability of Bayes was inlroduced,and the benchmark height that might be the aw^rage height of the left or the right sides of the V-gronve was acquired through the way ,and the interferences data of the strong faulae could be removed as possible as it could be combining with the depth of V-groove;At last,the left and the right side of＇ the Vgroove weld were fit separately with the least square method using the data which had been classified by the minimum error probability of Bayes and the center coordinate of the Vgrtoove weld could be captured well and truly.