A Novel Multi-Swarm Particle Swarm Optimization Algorithm Applied in Active Contour Model
Rui Li, Yirong Guo, Yujuan Xing, Ming Li · 2009
PSO (particle swarm optimization) algorithm provides a robust and efficient approach for searching for the object's concavities with the snake model.However, since single particle swarm optimization algorithm converges slowly and easily converges to local optima, it is not suitable well to be applied in active contour model directly. In this paper, a novel multi-swarm particle swarm optimization method was proposed to solve this problem. The proposed algorithm could expand the control point of the searching area and optimize convergence speed. It sets swarm for each control point and then every swarm search best point collaboratively through shared information, so it avoids the premature deficiency in traditional PSO algorithm. Compared our proposed algorithm with traditional algorithm, the experimental results showed that our method has superior performance than conventional snake model without spending extra time.