纺织学报 ›› 2010, Vol. 31 ›› Issue (11): 99-103.

• 染整与化学品 • 上一篇    下一篇

基于颜色空间与聚类分析相结合的袜带辨色分级方法

黄鹤玲;李霆;易积政;倪兵兵   

  1. 五邑大学信息学院
  • 收稿日期:2009-11-12 修回日期:2010-06-12 出版日期:2010-11-15 发布日期:2010-11-15
  • 通讯作者: 黄鹤玲

Color grading Method of stockings using Color space and FCM Clustering algorithm

HUANG Heling; LI Ting;YI Jizheng;NI Bingbing

  

  1. Department of Information, Wuyi University
  • Received:2009-11-12 Revised:2010-06-12 Online:2010-11-15 Published:2010-11-15

摘要:

针对丝卷检测中锦纶长丝染色均匀度测定的问题。采用CCD摄像头提取袜套样本图像,提出了应用商图像归一化的方法对袜套图像进行光源空间均匀性校正;应用多项式回归模型对摄像头提取的袜套样本图像进行色度特征化处理,得到的平均和最大色差分别为0.439和1.328 CIELAB色差单位;通过HSV颜色空间的色度百分率直方图,得到待检测袜套的颜色特征值,再由经验方程将其转化为聚类分析的输入矢量,最后应用聚类的方法对全体袜套的特征进行分析,得到袜带的染色分级区间及各段袜套的染色评级。结果表明,取得聚类过程中的参数最大迭代次数为100,终止误差限为1×10-5,模糊加权指数为2,得到聚类中心个数分别取2、3、4、5时的分类结果与人眼视觉比较,误差在0.125%以内,该误差远低于生产技术指标的允许误差,可用于对袜套颜色指标的检测和分级。

Abstract:

For the problem of inspecting the levelness of dyed polyamide filaments, sample images of stockings are extracted by CCD camera, and Quotient image normalization method is proposed for the correction of optical source uniformity of the extracted stocking images, whose color feature processing is carried out by using the polynomial regression model. After processing, their mean/maximum color differences are 0.439/1.328 CIELAB units. Then color features of each sample are obtained via adopting hue percentage histogram of HSV color space algorithm. These features are regarded as input vectors for Fuzzy C-means clustering algorithm (FCM) through empirical equation. Finally, the color levelness rating of the stockings are automatically created by means of FCM. The experiment shows that the maximum number of iterations is 100, the bound of settle-out error is 1×10-5 , the fuzzy weighting index is 2, and the error between the experimental value and eye viewing result is 0.125% as showed when the single digits in the clustering center are 2、3、4、5 respectively. This error is far less than the permit error of production and technical standard.

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!