基于行程长度纹理特征的焦炭显微图像分类算法  

Coke microscopic image classification algorithm based on run-length texture features

作　　者：王培珍[1] 缪家龙[1] 周可[1] 张代林[2] 

机构地区：[1]安徽工业大学电气信息学院,安徽马鞍山243002 [2]安徽工业大学煤的洁净转化与综合利用安徽省重点实验室,安徽马鞍山243002

出　　处：《煤炭学报》2012年第6期1051-1055,共5页Journal of China Coal Society

基　　金：国家自然科学基金资助项目(50874001);安徽工业大学研究生创新研究基金资助项目(2010020)

摘　　要：在分析焦炭显微图像各向异性光学组织中片状与纤维状显微图像特征的基础上,通过对纹理特征的差异性的研究,提出了一种基于行程长度纹理特征和支持向量机(Support Vector Ma-chine)的焦炭显微图像分类方法。该方法首先计算焦炭显微图像中4个方向上的行程长度矩阵,利用行程长度矩阵求得对图像纹理具有不同表征能力的纹理特征量,通过对各个特征量的数据分析,选取有效特征量组合作为分类器的训练向量,然后用支持向量机对实验样本进行分类。实验结果表明,该方法能够有效地识别出焦炭各向异性组织中纤维状、片状等不同光学组分。After analyzing texture characteristics of graphite and fibrous subclass in coke optical texture micrograpl~ classification algorithm, combining run-length textural features and Support Vector Machine （SVM）, was propose Firstly, the run-length matrix and a series of related texture features of coke optical texture micrograph were calcul ed, validities of these features for subclasses classifition were analyzed. Then Support Vector Machine based classi｜ was trained with those valid features and their combination. Experimental results show that with the proposed alt rithm, some subclasses among different optical textures of anisotropic, such as fibrous and graphite, can be classif more reasonably and effectively.

关 键 词：焦炭光学组织 纹理特征 行程长度 支持向量机 分类器 

分 类 号：TQ533.6[化学工程—煤化学工程]