本文针对舌色苔色的分类与识别,提出了一种DAG和决策树结合的方法。在训练SVM分类器的过程中,根据舌象样本中部分类别线性可分,而另一部分类别线性不可分的特点,采用了不同的核函数及其参数。实验结果表明,本文提出的算法在识别率和识别速度上都有一定提高。
A scheme based on DAG combining with DT is presented according to the characteristic of tongue manifestation. In the process of training classifiers, according to the characteristic of linear discrimination in the samples, different kernel functions and parameters are adopted in each two-class SVM. Theory and experiment result show that the rate and the speed of recognition with the method presented increase greatly.