Clustering-based image retrieval using fast exhaustive multi-resolution search algorithm

This paper presents a fast exhaustive multi-resolution search algorithm in a clustered image database. Prior to search process, the whole image data set is partitioned into a pre-defined number of clusters having similar feature contents. For a given query, the proposed algorithm first checks the lower bound of distances in each cluster, eliminating disqualified clusters. Next, it only examines the candidates in the surviving clusters through feature matching. Simulation results show that the proposed algorithm guarantees very rapid exhaustive search...