Abstract: Aiming to the problem with poor classification accuracy of present image classification methods because they fail to apply fully complementary advantages between various single features of images and redundant information exists in the extracted features, this paper presents an image classification method based on multi-feature and improved Support Vector Machine(SVM) ensemble algorithm. Comprehensive features describing fully image content are extracted; redundant information is removed by transforming extracted features with Principal Component Analysis(PCA). RBaggSVM classifier is applied for classification. Simulation experimental result shows that this method has higher accuracy and faster speed of image classification than similar methods.

Key words: multi-feature, Principal Component Analysis(PCA), Support Vector Machine(SVM) ensemble, PCA-RBaggSVM algorithm, image classification

摘要： 现有图像分类方法不能充分利用图像各单一特征之间的优势互补特性，提取的特征中存在大量冗余信息，从而导致图像分类精度不高。为此，提出一种基于多特征和改进支持向量机(SVM)集成的图像分类方法。该方法能提取全面描述图像内容的综合特征，采用主成分分析对所提取的特征进行变换，去除冗余信息，使用支持向量机的集成分类器RBaggSVM进行分类。仿真实验结果表明，与同类图像分类方法相比，该方法具有更高的图像分类精度和更快的分类速度。

关键词: 多特征, 主成分分析, 支持向量机集成, PCA-RBaggSVM算法, 图像分类

CLC Number: 

• TP393