交叉毛羽的凹点匹配分割算法

针对纺织行业纱线检测中对于毛羽重叠交叉问题涉及较少,难以准确检测的问题,提出一种基于凹点匹配的方法实现重叠交叉毛羽的分割。首先对纱线图像进行灰度化处理,然后采用形态学方法重建纱线图像,其次利用基于水平集的测地活动轮廓模型(GAC模型)检测纱线毛羽图像的边缘,对边界轮廓进行角点检测,通过设置夹角阈值来寻找纱线毛羽图像轮廓曲线角点中的凹点。最后通过重叠交叉毛羽区域产生的凹点进行匹配构造分割线,对重叠交叉毛羽进行分割。结果表明:该算法对于重叠交叉毛羽的2根分割准确率可达96.8%,对于3～5根重叠交叉毛羽分割准确率可达73.4%。

The cross hairiness segmentation algorithm based on the pits matching

Aiming at the problem of the difficulty in accurately detecting the overlap hairiness during the yarn inspection in the textile industry, a method based on pits matching was proposed to realize the segmentation of overlap or cross hairiness. First, the yarn image was grayed, then the morphological method was used to reconstruct the yarn image, and then the Geodesic Active Contour Model (GAC model) based on the level set was used to detect the edge of the yarn hairiness image. Second the angle between the boundary point and its successor and subsequent points was used to find the concave points on the contour of the yarn hairiness image. Finally, by matching the concave points generated by the cross hairs, the matching structure dividing line was drawn. The experimental results show that segmentation accuracy of the proposed algorithm is up to 96.8% for the two cross yarn hairiness, which is 73.4% for the three and more cross yarn hairiness.