棉花等级分类的预处理和特征提取  

Preprocessing and Feature Extracting of Cotton Classification

作　　者：刘立海[1] 王延平[1] 

机构地区：[1]武汉大学电子信息学院,武汉430072

出　　处：《数据采集与处理》2001年第2期145-149,共5页Journal of Data Acquisition and Processing

摘　　要：提出了棉花等级自动分类的实现框图 ,着重讨论了消除光线不均匀的预处理方法 ,提取了棉花图像的均方差变异系数、黑斑和黄斑三个特征参量。实验表明 ,这些特征能很好地表征棉花的等级。Cotton classification is to evaluate cotton′s quality, which influences the behaves of cotton individuals, cotton firms and mils. It is important to studying this issue. This paper presents a scheme of automatic classification for cotton based on standard cotton samples. It consists of three procedures: preprocessing, feature extracting and classifying. In the preprocessing we mainly focus on removing the non uniform rays which interfere feature extracting. Subtracting and dividing method is discussed. We extract three statistic features namely variance variation coefficient, dark spots and yellow spots. The first one stands for the uniformity of cotton, the second one denotes the whiteness and the last one reflects the quantity of bad spots (mostly are yelllow). Experiments show that these features work well for cotton grading.

关 键 词：预处理 棉花等级分类 特征提取 模式识别 图像处理 

分 类 号：F307.1[经济管理—产业经济] TP391.41[自动化与计算机技术—计算机应用技术]