A Sub-vector Weighting Scheme for Image Retrieval with Relevance Feedback

In this paper, a sub-vector weighting scheme is proposed for the case of small sample in image retrieval with rele-vance feedback. By partitioning a multi-dimensional visual feature vector to multiple sub-vectors, the singularity prob-lem caused by small sample can be avoided by the lower dimensionality of the sub-vectors. Then the optimal weight-ing can be performed on these sub-vectors respectively and the similarity scores obtained are combined as the final score to rank the database images. Experimental results demonstrated that the proposed weighting scheme can sig-nificantly improve the efficacy of image retrieval with rele-vance feedback