基于空间灰度共生矩阵木材纹理分类识别的研究

以10种木材纹理样本为对象,研究了木材纹理参数体系的建立方法,并进行了分类识别的仿真实验。首先,针对木材纹理特点并结合类别可分性判据,构造了适于描述木材的空间灰度共生矩阵,并在此基础上提取了木材的11个纹理特征参数。其次,借助相关性分析对参数进行了特征选择,进而建立了能直接与人的感官对应的木材纹理参数体系。最后,利用 BP 神经网络分类器对木材样本进行了分类识别研究,识别率为87.50%,验证了参数体系的有效性,表明用本文提出的纹理参数体系对木材进行分类识别是可行的。

Wood Texture Classification and Recognition Based on Spatial GLCM

This paper introduced the methods to build the parameter system of wood texture based on 10 categories of wood texture samples and the simulation tests of classification and recognition.First,the GLCM,which was suitable to describe wood texture,was established by the wood texture features and sort separability criterion.11 parameters of texture features were extract- ed from the GLCM.Then,feature selection of the parameters was done based on the correlation analysis.The parameter system of wood texture was buiht up,which could directly corresponded to human senses.Finally,the wood samples were selected and rec- ognized by BP neural network.The recognition ratio was 87.50%.The results verified the validity of the parameter system and in- dicated that the system was feasible in the selection and recognition of wood.