基于样本的棉花异性纤维识别算法

发布时间:

3O卷第4期 
陕西科技大学学报 
Journal of Shaanxi University of Science&Technology 
VoL 3O No.4 Aug.2012 
2O12年8月 
文章编号:1000—5811(2012)04—0110—04 
基于样本的棉花异性纤维识别算法 
李 娜 
(华北水利水电学院电力学院,河南郑州
450011) 
摘 要:异性纤维在棉花中虽然占得比重很小,但危害很大,直接影响着纺织品的质量.传统的 异纤检测识别算法多采取固定阂值法,但是由于检测对象是高速运动的棉花流,光照易产生波 动,采集的图像也会相应变化,因此误识别率就会增加.而有样本识别算法采集适量的分类样 本库,识别结果稳定,可以消除这一缺陷.分析棉花中异性纤维的图像特征,对重要图像特征进 
行增强,提取异纤的特征,识别算法引入欧几里德距离,采用K近邻分类,从而识别出异纤,并 进行仿真实验,实验结果表明该识别算法识别效果好、速度快. 关键词:异纤;识别;欧几里德距离;K近邻分类;算法 中图法分类号:TP751 
文献标识码:A 
Identification algorithm of foreign fibers in raw cotton 
base on sample database 
LI Na 
(Institute of Electric Power,North China University of Water Conservancy and Hydropower,Zhengzhou 450011.China) 
Abstract:Foreign fibers is a small proportion in cotton,but it seriously affect the quality of textile.The traditional method of fixed threshold is frequently used to identify foreign fibers 
n cotton,but detection object is high speed cotton flow,which easiy lead iumination to 
fluctuate.So the color of captured images will be affected accordingly,then misidentification 
possibility will be increased.However,the suitable amount of sample libraries are used in 
the identification algorithm of supervised classification,recognition result is stable,which e— 
liminate the defect.In this paper,the image character of foreign fibers in cotton is analysed
 
Further,important image features are enhanced by image processing,and foreign fibers characters are drawn.Euclidean distance and K—nearest neighbor classification are adopted in 
identification algorithm,and finally foreign fibers are identified

The results of simu1ation 
. 
experiments show that identification algorithm is simple,stable and efficient
Key words:foreign fibers;identification;Euclidean distance;K—nearest neighbor classifica 
tion;algorithm 
0 引言 
我国的纺织品便宜但是档次较低,与国际先进水平 相比,仍然存在较大差距.其中所用棉花中含有较 多异性纤维是导致原材料质量不高的重要原因之 
我国是纺织大国,而棉花是纺织工业最主要的 原材料之一,是关系国计民生的重要战略物资 . 

收稿日期:2O1 2-06—05 作者简介:李
娜(1 982),女,河南省濮阳县人,硕士研究生,研究方向:智能控制 

基于样本的棉花异性纤维识别算法

相关推荐