一种基于2DPCA的煤岩识别新方法  

A New Method of Coal Petrography Recognition Based on 2DPCA

作　　者：贾新泽[1] 杨慧贞[2] 段晋有[3] 田甜[4] 程永强[4] 

机构地区：[1]太原理工大学邮电中心,太原030024 [2]太原理工大学继续教育学院,太原030024 [3]太原理工大学国有资产管理处,太原030024 [4]太原理工大学信息工程学院,太原030024

出　　处：《太原理工大学学报》2011年第5期531-533,538,共4页Journal of Taiyuan University of Technology

基　　金：山西省自然科学基金资助项目(2010011019-1)

摘　　要：针对传统PCA方法用于煤岩识别常常面临图像维数高,直接计算数量大的问题,提出并实现了一种基于2DPCA的煤岩识别方法。这种识别方法是基于图像矩阵的主分量分析法,由于它的协方差矩阵可由原图像矩阵直接构建,因此2DPCA使用的协方差矩阵同传统PCA相比要小很多。实验结果表明,在训练样本数相同的情况下,2DPCA耗时仅占PCA总耗时的60%左右,并且随着训练样本的增多,2DPCA与PCA之间的耗时差会越来越大。识别率较PCA方法提高了近10%,图像SNR也由原来的4.53 dB提高到12.17 dB。2DPCA在速度方面表现优越,准确性方面也令人满意,有效的提高了煤岩识别的效率。Principal component analysis （PCA） method is a kind of one-dimensional feature extraction methods commonly used in image recognition. But the traditional PCA method used in coal recognition often faces with such issues： high dimensionality and large computation. This pa- per proposed and implemented a kind of coal recognition algorithm based on the 2DPCA . Its principal component analysis is based on image matrix, and its covariance matrix can be construc- ted directly from the image matrix, so the covariance matrix used by 2DPCA is much smaller than that used by PCA. The results show that in the case of the same number of training samples, the time consumption of 2DPCA accounted for about 60% that of PCA, and with the increase of training samples, the difference in the time consumption between 2DPCA and PCA increased. The recognition rate was improved by nearly 10% over PCA method, and the image SNR was al- so increased from 4. 53 dB to 12.17 dB. The speed and accuracy of 2DPCA method was satisfacto- ry, and the efficiency of coal recognition was improved effectively.

关 键 词：煤岩识别 PCA 2DPCA 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]