基于图像匹配的制衣裁片对横条方法

针对格子类衬衫的衣服裁片缝制时需要对条对横的基本要求,提出在机器视觉范围内制衣裁片对横条方法。此方法通过工业相机获取需要缝制的格子类裁片的图像。首先,运用阈值法分割出裁片,孔洞填充以及形态学方法使锯齿边缘消除,通过面积大小识别前幅后幅并进行校正;然后,根据纹理灰度变化梯度计算裁片最小格子单元,选取缝合边缘区域大小为6~8个最小纹理周期作为模板,匹配另一相应缝制区域裁片,模板及匹配区域同时包含非裁片区域。经实验对比了几种不同模板及匹配算法后,分析裁片对横条产生偏移的原因,提出基于灰度的互相关匹配的裁片对横条方法,该方法能够不偏移的对齐横条,且误差小于1像素。

Clothing pieces stripes alignment method based on image matching

The cut pieces of stripe garment are sewed according to the method of stripe matching, which is the basic requirement of making garment.Hence, the alignment method of cut pieces of garment based on image matching in the field of machine vision is proposed.The proposed method acquires image information of cut pieces to be sewed by industrial camera.First, cut-pieces are segmented by using threshold method, jagged edges are eliminated by using morphological methods, and front part and back part are identified by size.Then, gray gradient is used to calculate the smallest grid unit on pieces textured, the edge suture region includes 6~8 minimum texture cycles as a template to match the other pieces corresponding to the sewing area, and template and matching region is both containing non-pieces region.After compared several classic matching algorithms and templates in experiment and analyzed the reasons of bias when aligning the bar pieces skewed, the method of aligning the stripes of clothing pieces based on gray correlation matching is proposed.The method can align the stripes of clothing pieces without bias, and the error is less than 1 pixels.