Parallel relevance feedback for 3D model retrieval based on fast weighted-center particle swarm optimization

In this study, we present a parallel approach to relevance feedback based on similarity field modification that simultaneously considers all factors affecting the similarity field for 3D model retrieval. First, we present a novel unified mathematical model which formalizes the problem as an optimization problem with multiple objectives and constraints. Secondly, our approach optimizes all the parameters synchronously by treating all the modification operations of the similarity field equally. Thirdly, we improved the standard particle swarm optimization in two different ways. Finally, we present several experiments that show the advantages of our method over existing serial ones.