Abstract: To deal with the problem of extracting the weld seam from color image, a fast weld seam extraction method based on color curve analysis is proposed in this paper. It transforms the color space from RGB to YUV and analyzes the color curve on the V channel to obtain a suitable threshold value for binarization. Then the V channel image is binarized and after operations of de-noising with morphology and merging the separated blocks, the segmentation template is obtained. Finally, the template is applied to the original color image to extract the weld seam. Experimental results on the application of extracting the weld seam from color images show that the proposed method can deal with an image （800×600） within 0.3 s and the accuracy ratio is 90% when the overlapped ratio of the extraction window is more than 70%, which verifies the high efficiency and accuracy of the proposed method.

Key words: image segmentation, weld seam, threshold segmentation, YUV color space

摘要： 针对彩色图像中焊缝条纹提取问题，提出一种基于颜色曲线分析的彩色图像分割方法。该方法首先将原始彩色图像转换到YUV颜色空间，对V通道进行颜色曲线分析，得到一个合适的二值化阈值；接着利用该阈值对V通道图像进行二值化，经形态学去噪和合并离散块处理，得到图像分割模板；最后将其运用到原始图像上，实现对原始彩色图像的焊缝条纹提取。将该方法应用到彩色图像焊缝条纹提取中，对于800×600的彩色图像，平均处理时间约0.3 s，与人工标注结果比较，提取框重叠比例70%以上的占90%，验证了该方法的高效性和准确性。

关键词: 图像分割, 焊缝, 阈值分割, YUV颜色空间