改进蜂群算法的图像阈值分割方法

为快速高效地进行图像分割，针对人工蜂群算法存在的收敛速度慢、易陷入局部最优解等问题，提出了一种基于改进人工蜂群算法分割二维OTSU图像的新方法。通过对蜜源更新过程中向当前最优蜜源方向进行引导，可以加快算法的收敛速度；为避免算法陷入局部最优并加快收敛速度，在局部搜索过程中逐步缩减了搜索范围并加入了放弃机制；针对较大梯度值无意义的问题，限定了蜜源范围，以提高算法的效率。最后结合具有不同直方图分布的图像进行了实验，结果表明了算法稳健、高效、快速的特性。

Image threshold segmentation method based on improved artificial bee colony

In order to segment images exactly and quickly and for the problems of poor at convergence speed and easy fall to local best in artificial bee colony algorithm (ABC), the paper proposed a new method based on a improved ABC algorithm segmenting two dimensional Otsu images. In the nectar update procedure, it was guide the search direction to the current best nectar to speed up the convergence speed of the algorithm. In order to avoid the algorithm falling into a local best solution and accelerate the convergence speed, it reduced the search range gradually and implemented the abandoning mechanism in the local search procedure.Considering meaningless problems of a large gradient vaule, it limited the nectar range in nectar initialization and updating process. At last, experimental results on images with different histogram distribution show that the algorithm characteristics is robust, efficient and fast.