基于双层动态开窗V型焊缝快速识别方法

Fast recognition algorithm of V type groove seam based on double dynamic window

为实现焊缝的快速自动跟踪,提出一种双层动态开窗算法,根据序列图像的相似性,以V型焊缝前一幅图像提取出的三个特征角点为坐标,进行第一层动态开窗,截取出整幅图像ROI区域,对ROI区域预处理和二值化,再在二值化后的图像中进行第二层动态开窗,在小窗内进行Hough变换提取特征直线,得到特征直线相交构成的特征角点,由特征角点构成的三角形求解其重心作为焊缝特征点,明显减少了处理数据量。通过Matlab仿真实验表明,双层动态开窗后的一幅图片运算时间约为35 ms,满足一些对图像处理实时性较高的V型焊缝跟踪系统。

To achieve the fast automatic tracking of welding seam,a bi-layer dynamic window algorithm is proposed.According to the similarity of sequences of images,the first-layer dynamic window as the ROI is opened based on the coordinates of the three characteristic corners of the former V type of groove seam image.The preprocessing and binarization algorithm are applied to the ROI.Then the second-layer dynamic window is opened on the segmentation ROI.In the second-layer small window,the Hough transform is to extract characteristic straight lines which intersection points are the characteristic corners whose center of gravity is the features of the welding seam.The amount of image data processing is reduced obviously using the bi-layer dynamic window algorithm.Through Matlab simulation experiments show that the operation time of processing each picture is about 35 ms using the bi-layer dynamic window algorithm proposed and that this algorithm can meet some high real-time image processing of v-type seam tracking system.