| Particle swarm optimization algorithm is a global swarm intelligence-based heuristicoptimization algorithm. The basic idea of the algorithm is derived from the study of social birdssimplified model and simulate the behavior of birds, of which each individual as well as to use theirown intelligence community, continue to learn and adjust, and ultimately get an optimalsolution.The algorithm used for solving combinatorial optimization problems and nonlinearproblems. Because of its many advantages, such as less control parameters, easy to understand,easy to implement, fast convergence, less the required domain knowledge, so the algorithmattracted wide attention of many scholars when it have been proposed, and gradually become a newhotspot, and successfully applied to image segmentation, function optimization, pattern recognition,neural network training, etc.Image segmentation is the most basic and important part of image processing, imagesegmentation refers to an intensity, texture, shape and color of image to divided the image intovarious regions with different characteristics, and extracting the target region of interest.In this paper, the main work is as follows:(1) In order to solve the oscillation of particles in the vicinity of the optimal solution generatedwhen flying, the paper improved the algorithm formula, Based on flight time linear decreasingparticle swarm algorithm was proposed, reducing the oscillation phenomenon, improves theperformance of the algorithm.(2) PSO is easy to fall into local optima, and efficiency is not high. For this speed PSO basedon iterative equation is proposed, The algorithm is more concise, greatly simplifies the analysis andcontrol of the evolution of the particle, but also makes it easier for the algorithm to converge.(3) The flight time linearly decreasing PSO algorithm and Otsu algorithm combined, const itutea PSO-based flight time linear decreasing Otsu algorithm. Apply it in image segmentation, and theinertia weight has been improved after combining algorithm. Through experiments we can see thatthe algorithm running time shortened, reducing the number of iterations, operating efficiency is alsoimproved. |
