Adaptive weighted fusion: A novel fusion approach for image classification

Score fusion is a very competent fusion approach and weighted score fusion is the most preferable score fusion approach. To automatically set proper weights is the most important key of weighted score fusion and it seems that there are no truly adaptive weighted fusion approaches at present. In this paper we design a perfect adaptive weighted fusion approach, which automatically determines optimal weights and no any manual setting is needed. Though the proposed approach is very simple and quite easy to implement, it can obtain better performance than previous state-of-the-art approaches.