基于小生境粒子群算法的图像分割方法

Method for image segmentation based on niching particle swarm optimization

为了得到分割图像的最佳阈值,提出了一种基于小生境粒子群算法的图像分割方法。小生境粒子群算法通过划分小生境的方法,保持了物种的多样性,克服了粒子群算法容易陷入局部解,后期收敛速度慢的缺点,提高了算法的全局寻优能力。该方法基于最大类间方差阈值分割技术,用小生境粒子群算法对适应度函数进行优化,得到最佳阈值,并用该阈值对图像进行分割。实验结果表明,与最大类间方差法,基于基本粒子群算法的最大类间方差分割法相比,所提出的方法不仅能得到理想的分割结果,而且分割速度也得到了提高。

To determine the optimal thresholds in image segmentation,a new method based on niching particle swarm optimization is proposed in this paper.By the method of dividing niches,niching particle swarm optimization has kept the diversity of species,overcome the drawback of basic PSO,such as being subject to falling into local optimization and having the poor convergence speed,and so improved the ability of seeking global optima.The method uses maximum between-class variance（MV） technique,by the optimization of the niching particle swarm optimization object function,the optimal thresholds can be gotten,and the image by use of the thresholds can be segmented.Experimental results show that compared to maximum between-class variance technique,MV based the basic PSO algorithm,the proposed method can not only obtain ideal segmentation results,but also improve the speed greatly.