Majority voting re-ranking algorithm for content based-image retrieval

Abstract

We propose a new algorithm, known as Majority Voting Re-ranking Algorithm (MVRA), which re-ranks the first returned images answered by an image retrieval system. Since this algorithm proceeds to change the images rate before any visualizing to the user, it does not require any assistance. The algorithm has been experimented using the Wang database and the Google image engine and has been compared to other methods based on two clustering algorithms namely: HACM and K-means. The obtained results indicate the clear superiority of the proposed algorithm.