基于遗传相关向量机的图像分类技术

摘要

为了克服当前图像分类算法分类精度低等难题,提出基于遗传相关向量机(G-RVM)的图像分类方法,应用遗传算法对相关向量机参数进行优化.首先,采用主成分析法(PCA)提取图象特征;其次,利用训练集的数据,结合遗传算法获得相关向量机最优参数,得到优化的相关向量机图像分类模型;最后,采用Corel图像数据库中的图像作为实验数据.实验结果表明遗传相关向量机的图像分类方法比现有的图像分类方法有着更高的分类精度.因此,遗传相关向量机方法非常适合图像分类.%In order to solve the problem such as low classification accuracy in current classification methods, image classification technology based on genetic algorithm and relevance vector machine is presented in the paper. Genetic algorithm is applied to select the parameters of relevance vector machine. Firstly, principal components analysis (PCA) is used to extract image feature. Secondly, the parameters of relevance vector machine is obtained by the genetic algorithm and training data. Finally, the images in Corel image database are used to testify the classification performance of the proposed method. The testing results show that the classification accuracies of G-RVM are better than those of current classification methods. Therefore, G-RVM is very suitable for image classification.