一种图像分类的多特征vague融合模型*

针对图像分类中传统的特征融合方式所形成的巨大特征空间甚至维数灾难问题，提出了一种基于vague融合的图像分类模型。通过同时给出支持和反对的证据，运用vague集的真假隶属函数对图像分类中多特征分类器的分类结果进行决策融合，多特征分类器的分类结果得到优化和综合，从而获得更准确、更稳定的决策分类结果。实验结果表明，运用此决策融合模型是可行的，同时，图像分类准确率得到了明显提高。

Multi-feature vague fusion model for image classification

For traditional way of image classification, feature fusion scheme would decrease classificatory quality or result in other problems such as curse of dimensionality.This paper proposed a novel approach trying to integrate different features in image classification. Vague set for positive and negativeevidences was applied to analyze and optimize the decisions obtained by multi-classifiers. Through integrating two sides of multiple classification decisions, the classification was optimized and synthesized, thus the processing and the result would be both powerful and stable. Experimental results show that the performance of the classification is greatly improved.