Multi-model image retrieval method based on rough set inference and colour mutual information

Abstract

Searching interested images accurately, which has received considerable attention from researchers in the field like image processing, computer vision and multimedia systems in the last decades, is a challenging problem. In view of this, a multi-model image retrieval method based on rough set inference rules and colour mutual information is proposed. At first, inference rules of rough set-based image retrieval model which accomplishes initial retrieval based on semantic words is proposed in this paper. Furthermore, in order to overcome the problem that the simplicity of semantic annotation can cause vagueness of image retrieval information, colour mutual information-based image retrieval module is exploited in a multi-model way to realise precise retrieval by considering local feature variation. Experimental results show that the proposed solution can improve the efficiency of image retrieval effectively.