提出了一种基于焊缝ＣＣＤ图象模式特征的焊缝轨迹识别的新算法，其基本思想是将焊缝等分成为段，每段的灰度分布用特征向量来描述，利用前一段的特征向量来识别下一段的特征向量。从而实现焊缝的识别．这种算法具有对各种坡口快速、准确地识别并能自适应局部和整体噪声等特点，可用于机汛无辅助光源焊接焊缝轨迹预规划，并对实时焊缝轨迹进行了识别 A new algorithm of weld trajectory identification based on the characteristic of weld CCD image pattern is proposed. The basic idea of ​​the algorithm is to divide the weld into segments, and the grayscale distribution of each segment is described by the eigenvectors. By using the features of the previous paragraph Vector to identify the next paragraph of the eigenvector. In order to achieve the weld identification. The algorithm has the characteristics of fast and accurate identification of various bevels and adaptive local and global noise. It can be used for pre-planning of welding seam trajectory without auxiliary light source and identify the real-time weld trajectory