An Improved Otsu Multi-Threshold Image Segmentation Algorithm Based on Pigeon-Inspired Optimization

Threshold segmentation is a simple and effective method in the field of image segmentation which has the widest application domain. And the improvement of efficiency and precision of the threshold segmentation has received extensive attention and research. Inspired with the bio-inspired intelligent optimization, this paper proposes an Otsu multi-threshold segmentation based on pigeon-inspired optimization. The basic idea of this method is: the Otsu multi-threshold segmentation method is used to design the objective function, and the interclass variance function is used as the fitness function. The iterative optimization process is performed by the pigeon-inspired optimization. In this process, the fitness function is used as a criterion for the solution and corresponds to the coordinate of pigeon in the pigeon-inspired optimization. The best segmentation threshold group is obtained when the pigeon finds the global best position. This method converts the problem of finding the optimal solution into the solving problem of multidimensional variables and effectively optimizes the solution process. For the purpose of verifying the feasibility and segmentation accuracy of this method, the multiple segmentation parameters of several classical images of this method are compared with parameters of other classic algorithms such as particle swarm optimization and fireworks algorithm. The experiments show that the improved Otsu segmentation method based on pigeon-inspired optimization can effectively improve the speed of threshold solution, and the double operators ensures the accuracy of the segmentation. The method has the advantages of superior convergence and convenience of implementation. Simultaneously, the segmentation effect is ideal with this modus.