基于WQPSO算法的多阈值图像分割算法的研究

ALGORITHM OF MULTI THRESHOLD IMAGE SEGMENTATION BASED ON QUANTUM PARTICLE SWARM OPTIMIZATION WITH WAVELET MUTATION

引入基于量子行为的粒子群算法(QPSO)应用于图像分割。QPSO不仅参数个数少、随机性强,而且能覆盖所有解空间,但由于QPSO的后期局部搜索能力较弱,因此提出一种基于小波变异的量子粒子群优化算法(WQPSO)以增强其局部搜索能力,保证算法的全局收敛性。把图像分割看成一个最优化问题,以最大类间方差法(OTSU)为例,对比了WQPSO、标准粒子群算法(PSO)和QPSO在阈值处理中的性能,实验结果表明WQPSO完全满足实时系统精确度和准确性的要求,具有无可比拟的图像分割效果。

Quantum-behaved Particle Swarm Optimization （QPSO） is applied to image segmentation. QPSO has few parameters, its randomicity is strong, and covers all the solution space. Because of the weak local searching ability in QPSO＇ s later stage, a wavelet mutation based QPSO, WQPSO is introduced to enhance local searching ability and guarantee its global convergence. In this paper,image segmentation is considered as an optimization problem. Taking the maximum between-class variance （OTSU） as an example, the performances in threshold computation are compared among WQPSO, the standard particle swarm optimization （PSO） and QPSO, and the experiment results demonstrate that WQPSO can completely satisfy the accuracy and correctness requirement of the real-time system and provide an incomparable effect on image segmentation.