基于JSEG算法的纺织品印花图像分割  

Segmenting color region of textile printing pattern image based on algorithem of JSEG

作　　者：李鹏飞[1] 王刚[1] 景军锋[1] 焦珂[2] 

机构地区：[1]西安工程大学电子信息学院,陕西西安710048 [2]西安机电信息研究所,陕西西安710065

出　　处：《纺织学报》2010年第5期137-140,共4页Journal of Textile Research

基　　金：中国纺织工业协会重点研究项目(2008041)

摘　　要：针对机器视觉对纺织品印花精度的检测问题,提出了利用JSEG算法对印花的纺织品图像进行分割。主要通过对图像的滤波,颜色的矢量化和区域的增长与合并,完成图像区域的分割。通过实验给出了颜色量化阈值和区域合并阈值对图像分割的影响以及选取的方法,最终给出该算法对印花纺织品图像的分割效果图,表明该算法利用颜色信息和区域信息,对包含纹理信息的彩色印花图像有良好的分割效果,并且对由日光引起的一些干扰有一定的抑制作用。In view to the problem of accurate detection of printed patterns of textiles based on the machine vision,this paper presented a method of applying the JSEG algorithm to the image segmentation of the textile printing pattern image,and completion of image region segmentation included image filtering,color vector,regional growth,and merger.The impact of the threshold value of color quantization and the threshold value of region merging on image segmentation and method selection was investigated by experiments.Finally,the results of image segmentation of textile printing pattern by using this algorithm were demonstrated,indicating that the algorithm which uses color information and regional information is efficient in image segmentation of weave information contained textile printing pattern,and shows certain inhibiting effect to some interference caused by the sun light.

关 键 词：纺织品印花图像 图像分割 区域生长 同态组滤波 JSEG 

分 类 号：TS194.33[轻工技术与工程—纺织化学与染整工程]