基于LBP-DEELM的木材纹理分类算法

Wood Texture Classification Algorithm Based on LBP-DEELM

为解决传统木材纹理分类的准确率低且难度大的问题,提出了一种基于LBP-DEELM(局部二值-差分演化优化极限学习机)模型的木材纹理分类算法。在阐述局部二值算子(LBP)和差分演化优化极限学习机(DEELM)算法的基础上,使用均匀旋转不变的LBP模式提取纹理的特征值,结合差分演化算法进行极限学习机优化,通过训练得到每类纹理所对应的分类器模型参数,构造分类器,实现了对木材纹理准确高效的分类。实验结果表明,相比于BP神经网络,SVM支持向量机等分类算法,该模型的实验误差率为2%左右,准确率高,实用性强。

In order to deal with the issue of low categorization accuracy and tough calamity,a wood texture classification algorithm based on LBP- DEELM was proposed. On the basic of LBP operator and DEELM algorithm,combining the rotation invariant properties and original LBP operator,extracting the texture eigenvalues,combine the differential evolution algorithm to improve the efficiency of extreme learning machine,then get the parameters through training of different wood texture,establish a classifier,which achieves accurate and efficient wood texture classification. The experimental results show that the model error rate is about 2%,has higher accuracy and practicality than BP Neutral Networks,Support Vector Machine.