基于KCPA提取特征和RVM的图像分类

Image Classification Based on KCPA Feature Extraction and RVM

为了得到更理想的图像分类结果,提高图像分类的效率,提出一种核主成分分析与相关向量机(RVM)相融合的图像分类算法.首先采集大量图像,建立图像数据库,并提取图像特征;然后采用核主成分分析对图像进行选择和降维,减少图像特征数量,消除作用较小的特征;最后通过相关向量机的训练构建图像分类器.采用3个图像数据集进行图像分类实验,实验结果表明,对于3种标准图像数据库的图像,该算法的图像分类正确率大于95%,远高于其他算法的图像分类正确率,且图像分类速度可以满足图像的实际应用要求.

In order to get better result of image classification and improve the efficiency of image classification,we proposed image classification algorithm based on kernel principal component analysis and relevance vector machine（RVM）.Firstly,a large number of images were collected to establish the image database,and features were extracted.Secondly,kernel principal component analysis was used to select features and reduce dimension of image to reduce the number of image features and eliminate some small features.Finally,image classifier was constructed by training of relevance vector machine,and 3 kinds of standard image databases were used to perform image classification experiments.The experimental results show that,for 3 kinds of standard image databases,the image classification accuracies of the proposed algorithm are more than 95%,which is much higher than that of other classification algorithms,and the image classification speed can meet practical requirements of the image.