Context re-ranking in sketch based image retrieval

Abstract

Image retrieval based on the sketch-based descriptor is focused in this paper. The retrieval operation based on the sha pe details defined in a sketch input is used for image region descriptor and a distance based mapping approach is developed for image retrieval in a database system. The search overhead, decision accuracy and feature representation is constraint to such sketch based approach, hence in this paper, a context re-ranking model based on feedback modeling is proposed. The approach has an advantage of faster retrieval performance compared to the conventional retrieval system. The validation is made with the simulation result developed for the proposed approach over the conventional benchmark approach.