基于增强布谷鸟搜索的图像分割算法

Image Segmentation Algorithm Based on Enhanced Cuckoo Search Algorithm

针对传统图像分割算法对复杂、含噪图像的性能较差的问题,提出一种基于增强布谷鸟搜索与snake模型的医学图像分割算法。首先,为布谷鸟搜索算法引入两点增强策略:鸟蛋放弃策略与优质鸟蛋之间的信息交互策略,由此加速布谷鸟搜索的收敛过程;然后,设计了1个两阶段的图像分割算法,第一阶段设计了局部搜索窗口对snake的所有控制点进行约束,并通过布谷鸟搜索使snake能量最小化,第二阶段通过优质鸟蛋之间的信息交互策略,获得布谷鸟搜索的新控制点。基于真实MRI图像的实验结果显示,本算法对复杂、含噪图像的分割性能优于其他snake模型的分割算法。

Concerning the problem that the traditional image segmentation algorithms show poor performance for the complex and noisy images, an enhanced cuckoo search algorithm and snake model based medical image segmentation algorithm is proposed. Firstly, two enhanced strategies are introduced to the cuckoo search algorithm： abandoned eggs strategy and information exchange between high quality eggs, which speedups the convergence procedure of cuckoo search; then, a two-phase image segmentation algorithm is designed, the first phase, the local search window is designed to constrain the control points of snake, and the snake energy is minimized by cuckoo search, the second phase, the new control point is generated by information exchange between high quality eggs. Experimental results based on realistic MR images show that the proposed algorithm performs better segmentation results than the other segmentation algorithms based on snake model.