Applying Particle Swarm Optimization to Transfer Function Specification for Direct Volume Rendering

Transfer function (TF) specification is an important research issue in direct volume rendering (DVR). In this paper, the problem has been developed as a parameter optimization model and a modern evolutionary algorithm, the particle swarm optimization (PSO), has been applied to solve it effectively. Initial solutions are formed and encoded into particles of PSO, the particles fly intelligently in the search space to achieve the best sequence. We realize the implementation of the algorithm in automatic and interactive approaches, and modify updating strategy emphasizing the user requirement in the interactive method. Experiments results confirm the performance and efficiency of the automatic implement of origin algorithm with a comparison to genetic algorithm (GA) which has been applying to this field previously. A contrast has been made between the conventional and modified PSO algorithm and GA in interactive implement to verify the feasibility and applicability of modified PSO algorithm.