根据织物表面图像来自动识别组织结构参数是纹织CAD的一个重要研究内容。为解决在扫描过程中织物图像不可避免的倾斜现象,提出了一种快速的基于Hough变换的织物图像倾斜检测算法。为减少运算量,此算法首先提取图像梯度作为纬纱走向信息;然后运用层次Hough变换来检测倾斜角度,并获得了满意的检测精度;最后根据倾斜检测结果,采用一种新的与倾斜无关的纬密识别算法,通过提取倾斜角处的投影轮廓线来得到纬密排列规律,并计算出纬纱密度。实验结果表明,该算法用于结构图像倾斜检测和纬纱密度识别,可获得大于88%的检测准确率,纬度识别倾斜误差可控制在2°以内,可见具有较高的准确率和较好的实用性。

Automatic identification of structural characteristics is an important research area for computer aided design of woven fabric.Since the skew is inevitable during fabric scanning,a fast skew detection algorithm based on Hough transform for the fabric surface image is proposed.In order to reduce compute complexity,the information of weft direction is extracted from the gradient of fabric image firstly.Then,hierarchical Hough transform is used to estimate the skew angle of fabric image with satisfactory precision.Finally,based upon the skew detection process,a novel algorithm to identifying the weft density is introduced,which is independent of the image skew.The projection profile on skew angle is used to determine the arrangement of weft yarns,and the density of weft is easily computed.The experimental results show the practicability and efficiency of our method.