摘要
本文用主分量分析法分析了木材纹理的14个灰度共生矩阵特征参数,从中提取了4个综合参数,并分别统计了采用这两套特征参数,最近邻分类器,K近邻分类器和神经网络分类器对木材样本分类正确率,结果表明采用主分量分析提取的综合参数不仅能减少数据量,而且获得了较高的分类精度。
This paper analyzed 14 wood texture＇s feature parameters of GLCM, extracted 4 integrated parameters, and noted the correct classifying ratio of wood samples with nearest neighbor classifier, K neighbor classifier and neural network classifier by using two sets of feature parameters. The results showed that using integrated parameters extracted by PCA could not only reduce data quantity, but also acquire higher classifying precision.
出处
《森林工程》
2006年第6期14-16,共3页
Forest Engineering
基金
黑龙江省自然科学基金项目(C2004-03)及(C0308)
哈尔滨市自然科学基金项目(2004AFXXJ020)
