基于Radon变换的织物倾斜和纹理方向检测

Detection of fabric skew and texture direction based on Radon transform

为减少织物倾斜对自动检测的影响,采用Radon变换对织物图像中织物倾斜和纹理方向进行检测。先利用计算机模拟生成与纱线粗细相同和间隔一致的平纹织物,对该图像进行Radon投影变换,分析经纬纱和纹理方向与Radon变换投影轮廓曲线峰值的对应关系,得出图像倾斜角和投影角之间的关系。之后对10个织物图像中织物倾斜进行检测,分析织物的经纱方向、纬纱方向、纹理方向、经纬向夹角。通过与人工测试数据比较发现,基于Radon变换的测试方法对织物图像的倾斜和纹理方向检测更具有优越性。

In order to reduce the impact of fabric skew on automatic detection,fabric skew and texture direction in a fabric image were detected by Radon transform. Firstly the image plain fabric with same parameter value of the yarn diameter and the yarn space was simulated by computer programming. Then the simulated image was transformed by Radon projection. The relationship between direction of warp yarn and weft yarn and texture and projection curve peaks was analyzed. Finally,for images of ten fabric samples,the direction of warp yarn and weft yarn,texture direction,the angle between warp yarn and weft yarn were analyzed. Compared with manual test data,it is found that the method based on Radon transform is more favorable to the detection of the skew and texture direction of the fabric image.