基于均匀颜色空间的木材分类研究  

Research on wood classification using uniform color space

作　　者：王克奇[1] 杨少春[1] 戴天虹[1] 白雪冰[1] 

机构地区：[1]东北林业大学机电工程学院,黑龙江哈尔滨150040

出　　处：《计算机工程与设计》2008年第7期1780-1784,共5页Computer Engineering and Design

基　　金：黑龙江省自然科学基金项目(C2004-03);哈尔滨市自然科学基金项目(2004AFXXJ020)

摘　　要：木材材色在各种颜色空间中的分布范围都比较窄,而利用CIE1976L*a*b*均匀颜色空间的等距性和色差高分辨力的特点可以很好地对其进行测量与辨别。在L*a*b*颜色空间中,提取了东北常见的5种树种的径切和弦切木材图像的颜色特征,然后采用模拟退火算法对这些特征进行了选择,并结合BP神经网络和-近邻两种分类方法进行了仿真试验,得到了满意的分类结果。The wood color range over a narrow area in all kinds of color space, but it is much suitable to measure and distinguish them by using CIE1976 L^＊a^＊b^＊ uniform color space as it has the characteristic of isometry and high resolving power. The color features of five tree species are extracted in L^＊a^＊b^＊ color space, and some of them are then chosen with Simulated Annealing. With BP neural network and k near neighbor, the simulation experimentation of classification is done and the result is satisfying

关 键 词：木材材色 L^＊a^＊b^＊颜色空间 色差 模拟退火算法 特征选择 神经网络 κ-近邻 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]