融合改进人工蜂群和K均值聚类的图像分割

Image segmentation algorithm based on improved artificial bee colony and K-mean clustering

针对人工蜂群优化的K均值算法易陷入局部最优、搜索精度不够、分割图像不够细致等问题,本文融合自适应人工蜂群和K均值聚类,提出了一种新的图像分割算法。算法首先利用距离最大最小乘积对种群进行初始化;其次采用自适应搜索参数动态调整邻域搜索范围,使人工蜂群算法快速收敛于全局最优;然后将人工蜂群输出的所有蜜源进行K均值聚类,克服K均值聚类结果对初始聚类中心的依赖,再将聚类划分结果进行Powell局部搜索,加快算法收敛的速度,将得到的新聚类中心更新蜂群中蜜源位置。最后,将本文算法与其他两种同类分割算法进行试验对比。实验结果表明:与其他两种算法相比,本文提出的分割算法在保证运行时间的前提下,分割准确率比其他两种算法分别至少提高了3.5%和4.8%,表现出了较高的分割质量。

In order to overcome the artificial colony optimization k-means which be fallen into local op- timum easily, converged slowly, segmented roughly and other issues, a new image segmentation al- gorithm is proposed based on adaptive artificial bee colony and K-mean clustering. First, the popula- tion is initialized by the maximum and minimum product; Secondly, adaptive search parameters are used to adjust neighborhood search scope dynamically, that makes artificial bee colony algorithm quickly converge to global optimal and achieve a more optimal solution; Then, all neetaries will be clustered by K-mean to the dependence of clustering result on the initial center, and then clustering results are divided into Powell local search, which accelerate the algorithm convergence speed, that will receive a new clustering center update colony of nectar source location. Finally, the proposed algo- rithm is compared with the other two algorithms. The experimental results show that compared with the other two algorithms, the segmentation algorithm proposed in this paper can improve the segmen- tation accuracy by at least 3.50% and 4.8%, respectively, under the premise of guaranteeing therunning time, showing a higher segmentation quality.