Fusing global and regional features for image classification

This paper presents a novel approach to image classification based on the fusion of global and regional features, which are helpful to describe image semantics to classification, in which vague sets for positive and negative evidences is applied to analyze and optimize the decisions obtained by multi-classifiers. Through integrating two sides of multiple classification decisions, the classification is optimized and synthesized, thus the processing and results will be both powerful and stable. Experimental results show that the performance of the classification is greatly improved.