采用纹理平滑的印花织物图案轮廓提取  

Graphic contour extraction for printed fabric based on texture smoothing

作　　者：向军 张杰[3] 潘如如[1,2] 韩要宾 张继东[1,2] 高卫东[1,2] 

机构地区：[1]生态纺织教育部重点实验室(江南大学),江苏无锡214122 [2]江南大学纺织服装学院,江苏无锡214122 [3]香港理工大学纺织及制衣学系,香港九龙999077

出　　处：《纺织学报》2017年第11期162-167,共6页Journal of Textile Research

基　　金：国家自然科学基金青年基金项目(61202310);教育部博士点基金项目(20120093130001);霍英东基金项目(141071);2014江苏省研究生创新计划(KYLX_1132)

摘　　要：为丰富印花产品的种类,提出了一种可有效提取出印花织物中图案轮廓的方法,可提取出高精度的印花图案边缘轮廓。首先对图像进行平滑,通过控制平滑程度参数和空间尺度参数,平滑掉织物图像中的纹理结构。再用Canny边缘检测算子检测图案边缘,Canny分割的判别阈值采用默认自动选择阈值就能成功提取织物上图案的轮廓,分割后的图像轮廓清晰,边缘连续,并且能分割出印花织物图像中的细小结构。通过实验证明,在RGB颜色空间对图像平滑和边缘提取的效果要优于其他颜色空间,比较了其他边缘检测算子分割印花织物图案的效果,结果证明Canny算子分割效果最好。In order to enrich printed products varieties,a method was put forward to effectively extract the contours of the printed fabric with high precision. Firstly,in order to smooth the image,by controlling the degree of smooth parameters and space scale,it could smoothen the fabric texture and structure of the image. It could successfully segment the pattern on the fabric when edge detected by Canny edge detection operator,and Canny discriminant threshold segmentation selected the default automatic threshold choice.The segmented image had clear contour,continuous edge and could segment smaller structure in the images of the printed fabric. Experiments prove that in RGB color space image smoothing and segmentation effect are superior to the other color space. Compared to the other edge detection operator splitting pattern effect of printing fabrics, the results prove that Canny operator segmentation effect is the best, and the segmentation results can be directly used in the production of printed fabric.

关 键 词：印花织物 图像平滑 边缘提取 图案分割 CANNY算子 

分 类 号：TS194.9[轻工技术与工程—纺织化学与染整工程]