均衡策略粒子群算法在图像分割中的应用
Multi-level image segmentation based on an improved particle swarm optimization with an equilibrium strategy

针对图像阈值分割方法由于其穷举的性质使分割时间随着阈值数目的增加而无法满足图像处理的要求的问题，提出一种基于均衡策略的粒子群进化算法来缩短分割的时间，改进的算法在粒子运行过程中引入均衡因子以增强个体获得较大搜索能力的可能性，确保它能进行有效的全局搜索；此外，为了增强算法的局部搜索能力，在群体进化方向中引入一个扰动因子，从而使得个体能够在该方向上获得更多局部搜索机会。基于熵准则的Kapur用来作为验证提出的算法优劣。标准测试函数和标准图像实验结果表明，提出的算法相比于其它比较算法而言，在寻优能力和收敛速度上获得了更好的成绩。
The computation time of some Multi-level threshold segmentation techniques needs were too long to bear, and it grew exponentially with the number of thresholds increased. This paper proposed a particle swarm optimization with an equilibrium strategy for shorting its computation time. First, during iterations, a balance operator for individuals to have more chances to search in the global area was introduced. Furthermore, for enhancing the local search ability of our proposed algorithm, a disturbance operator was also been introduced in this paper which enabled the individual had more opportunities to make a precise search. The improved algorithm enables particles had more chances to jump out of a local area for enhancing their global search ability. Meanwhile, a valuable point to guide the search direction of the particles was introduced. Then it accelerated the convergence rate of the improved algorithm. Kapur method was used in this paper to test the performance of the proposed method. Experimental results showed that our proposed algorithm showed more power and fast search ability when compared with the other population-based algorithms.