多分类光滑支持向量机及Gabor滤波应用研究

Research of Multi-Class Smooth SVM and Gabor Filter Application

在传统SVM的分类求解算法中,严格凸的无约束最优化问题中单变量函数x+是不可微的。将原来不可微的模型变为可微的模型,可以使用通常的最优化的算法进行求解。Gabor滤波器采用不同方向、不同尺度,对烘焙面包切片区域灰度图像直接进行小波变换,用能量均值"和均方差#来表示灰度图像的纹理特征。针对样本数少的特点,提出使用多分类光滑支持向量机的方法实现对烘焙面包品质的快速分类。实验结果表明了该方法的实用性和可行性。

In traditional SVM solution algorithms, objective function is a strictly convex unconstrained optimization problem, but is not differentiable due to x. The undifferential model is converted into a differential model which could use the most used optimization algorithms. Gray images of the field of view are transformed with different orientations and scales by Gabor filters. The texture features are denoted by the energy standard value μ and standard variance σ. A multi-class Smooth Support Vector Machine is proposed to implement the quick classification of baking bread according to small amount samples. The results indicate the practicability and feasibility of the method.