원문정보
초록
영어
The particle swarm optimization (PSO) is a new swarm intelligence technique inspired by
social behavior of bird flocking. In this paper, the optimal multi-objective particle swarm
optimization (OMOPSO) is presented. Since the parameters determine the optimization
performance of the algorithm, the uniform design is introduced to obtain the optimal
combination of the parameters. Additionally, a new crowding operator is used to improve the
distribution of nondominated solutions, and ε-dominance is used to fix the size of the set of
final solutions. OMOPSO is applied to optimize the parameters of blind color image fusion.
First the model of blind color image fusion in YUV color space is established, and then the
proper evaluation metrics without the reference image are given, in which a new metrics of
conditional mutual information is proposed. Experimental results indicate that the method of
blind color image fusion based on OMOPSO realizes the Pareto optimal blind color image
fusion.
목차
1. Introduction
2. OMOPSO Algorithm
2.1. OMOPSO Flow
2.2. Repository Control
2.3. ε Dominance
2.4. Adaptive Inertia Weight
2.5. Crowding Distance
2.6. Adaptive Mutation
2.7. Uniform Design of Parameters
3. Blind Color Image Fusion in YUV Space
4. Evaluation of Blind Color Image Fusion
4.1. Gradient
4.2. Entropy
4.3. Conditional Mutual Information
5. Experiments
5.1. Uniform Design of the Parameters
5.2. Comparison among Different Fusion Schemes
6. Conclusions
7. References