RGB空间下基于C-means颜色聚类的印花布图像分割方法研究  

Print Fabric Image Segmentation Based on Fuzzy C-means Color Clustering in RGB

作　　者：王芳[1] 

机构地区：[1]浙江理工大学信电学院,浙江杭州310018

出　　处：《工业控制计算机》2011年第3期4-5,7,共3页Industrial Control Computer

摘　　要：针对印花瑕疵检测的关键技术问题,主要围绕颜色度量和彩色图像分割两个方面展开研究工作。提出一种新的颜色度量公式,实现直接利用聚类分析算法在RGB颜色空间中对图像进行分割。同时还提出一种基于距离修正的模糊C均值聚类算法,提高了算法的抗噪声性能。实验研究结果表明,所提出的算法能成功有效地对印花图像的颜色特征进行提取与分割,并且抗噪声性能也比较强。Focus at the key problem of print fabric flaw measurement,the paper is devoted to studying image color scale and segmentation method.To segment image by clustering directly in the RGB color space,it propose a new color scale method.To improve the anti-noise robustness for C-means clustering method,this paper propose a method to modify the membership function in the objective function and change the distance metric.The experimental results show that the proposed method can extraction print fabric feature effectively,and the anti-noise robustness is good.

关 键 词：印花布 瑕疵检测 RGB颜色空间 模糊C均值聚类 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]