一种识别薄板对接接头的视觉方法

A Vision System for Weldline Recogniton of Thin Metal Plates

对薄板紧密对接接头的识别问题进行了深入的分析，提出了由照明、自适应局部特征提取、全局特征识别和验证融合在一起的视觉识别方法。它克服了电弧光的干扰，识别结果可靠，可以有效地解决薄板紧密对接接头识别这一难题。

In order to solve the problem of tight-butt weld seam, a method based on active lighting and passive vision is put forward to recognize this type of seam. The active lighting makes the obtained images clear and stable. The global feature recognition erables the local feature points on the same line to cluster around one point in transformed coordinate system. It also changes the detection of straight line in the image into the search for the maximum in a one-dimension array, thus simplifies the identificaticn process of weldline. Thus the recogniton result becomes more reliable and stable. The limits checking and the final verification increases the reliability of misalignment or rolled-edge in welding joint, and the vision system has correctly recognized the tight butt joints of cold-rolled steel, hot-rolled steel and aluminum plates with sheared edges.