基于两方向二维主成分分析木材识别的研究  

Study on Timber Recognition Based on Two-oriental and Two-dimensional Principal Component Analysis

作　　者：何涛[1,2] 张世禄[1] 汪杭军[2] 

机构地区：[1]西华师范大学计算机学院,四川南充637002 [2]浙江林学院信息工程学院,浙江临安311300

出　　处：《林业机械与木工设备》2009年第10期12-14,共3页Forestry Machinery & Woodworking Equipment

基　　金：浙江省科技计划项目(2008C21087)

摘　　要：两个方向的费希尔主成分分析方法(2D)2FPCA结合了二维主成分分析方法(2DPCA)和二维费希尔线性方法(2DFLD)的特点,很好地解决2DPCA特征提取时比传统PCA需要更多系数来表达图像信息的问题。根据木材体视图受光照影响及同一树种的样本图片之间差别较大等特点,适当增加了识别的类内散布矩阵从而提高了木材的识别率。(2D)2FPCA为木材的智能识别提供了一条新途径。Two-oriental Fisher Principal Component Analysis （2D）^2FPCA combines the features of both Two-dimensional Principal Component Analysis （2DPCA） and Two-dimensional Fisher Linear Diseriminant （2DFLD）. It can resolve the problem which needs more coefficients to express image information using 2DPCA, compared with traditional PCA. According to the features that timber body views are affected by light and the large differences between the sample pictures of the same species of timber, within-class scatter matrix intended for recognition is added appropriately in order to increase timber recognition rate. （ 2D ）^2FPCA provides a new method for intelligent recognition of timber.

关 键 词：(2D)2FPCA 2DPCA 2DFLD 木材体视图 识别率 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]