基于复小波和支持向量机的纹理分类法*

针对图像纹理分类问题,提出了一种将二元树复小波变换与支持向量机相结合的分类方法,通过二元树复小波变换对纹理图像进行四层分解,提取各子频带小波系数模的均值和标准方差组成特征向量,利用支持向量机作为分类器实现纹理图像分类。对20类Brodatz纹理图像的分类实验表明,提出的方法具有较高的分类精度,在有限训练样本的情况下比传统的分类算法平均正确率有10%左右的提高,体现了该方法的有效性和良好的泛化能力。

Method of texture classification using dual tree complex wavelet transform and support vector machines

A novel method for texture image classification was proposed by using dual-tree complex wavelets transform and support vector machines.The dual-tree complex wavelets transform was used to decompose texture image with four levels,feature vector was generated by computing mean and standard deviation from coefficients of individual wavelet subbands.This feature vector was first used for training and later on for tesing the support vector machine classifier.The experimental setup consists of twenty texture images from the Brodatz image database,results of experimentent indicate that the presented method provide superior texture classification accuracy over other methods under the condition of limited training samples,and show the validity and the best generalization ability.