A multilevel threshold segmentation technique using self-adaptive Cuckoo search algorithm

Image segmented by the standard cuckoo search algorithm(CS) has large computational complexity and is easy to fall into the local minimum solution under slow convergence rate. A self-adaptive CS algorithm(SCS) is adopted for multi-threshold image segmentation in Otsu method to solve these problems. In order to improve the local search performance as much as possible and simultaneously keep the strong global search effect, a new inertia weight is introduced and the parameter which is called step size in the cuckoo search algorithm is changed to be dynamic, adaptively changing with the search process to improve the local search ability and the speed of convergence. The feasibility of recommended approach has been tested on several color images. Experimental results show the proposed algorithm has better segmentation quality, better segmentation results and faster convergence speed compared with a modified cuckoo search (MCS) algorithm.