Weight Optimization of Image Retrieval Based on Particle Swarm Optimization Algorithm

Block method can overcome the disadvantage of global color histogram image retrieval algorithm, but how to efficiently and accurately set the weights of blocks is an important issue in research area. This paper proposes a new approach for block color histogram image retrieval based on particle swarm optimization (PSO) algorithm, it converts the sub-block weight setting into optimization problem, and then uses the PSO for optimal solution to enhance search results. Experimental results show that both the recall and the precision are improved effectively, moreover, the proposed approach is able to find the best combination of block weight for different resolution images.