Re-weighting relevance feedback image retrieval algorithm based on particle swarm optimization

Aiming at the inflexible re-weighting problem of relevance feedback (RF) in image retrieval, a re-weighting relevance feedback method utilizing particle swarm optimization (PSORW-RF) is proposed. Firstly, initialize feature weightings randomly, then use the variances of the positive and negative feedback samples´ features as study principle, utilize particle swarm optimization (PSO) algorithm to optimize weightings according to user´s retrieval requirement, and obtain retrieval results at last. Experiments show that the proposed algorithm is validity.