Image Segmentation Research Based On Improved Particle Swarm Optimization Algorithm And 3d-Otsu

Image segmentation is the basic component in the process of digital image processing technology, also next to the image pattern recognition, analysis and processing. So the efficiency and accuracy of image segmentation is critical for follow-up work. Otsu, which is one of Threshold method, is simple, feasible and stable performance, become a core technology in image segmentation. But it will take a long time to find a more accurate optimal threshold combination in the whole gray area of a multi-modal image, which seriously hampered the application of multidimensional Otsu method. In order to get a relatively good segmentation results and reduce the integral time, this paper introduce particle swarm algorithm fusion Otsu method, and then take measures for improvement of particle swarm optimization(PSO).Related work and innovation point basically has the following several aspects.Aiming at the shortcomings of the basic particle swarm algorithm in prophase rapidity, initialized using chaotic mapping equation of particle population, enhance the uniformity of the initial distribution of the particle swarm, guarantee the diversity of initial population. Introducing the concept of cloud model in particle swarm optimization algorithm, the particle group in the iteration process is divided into three parts, respectively, of the inertia weight by different ways of operation. Using normal cloud generator to optimize their inertia weight with the part of the mean particles, make in the process of iteration for cloud adaptive adjustment, which can improve the convergence speed and convergence precision of the algorithm.In late iteration of the algorithm, The particle population must be variable in order to avoid the population into a local optimum. When the population gathered through highly, to improve the precision of the solution, we can use the normal cloud generator on the better kind of particles search in the neighborhood of the optimal particle; Using chaotic algorithm to disturbance the particles on the poor kind of particles, to enhance the diversity of population and avoid stagnation or algorithm into local optimum. Simulation experiment is used on the improved particle swarm optimization algorithm with the standard function, the results show that this algorithm is better than the basic particle swarm optimization algorithm and the chaotic particle swarm optimization algorithm on the convergence speed and accuracy.The between-cluster variance method only using the gray information of images, a one-dimensional Otsu method is extended to the 2d and 3d, and then weighted the neighborhood value to improve the image segmentation processing of edges and details. In order to decrease the amount of calculation of the Otsu method, a mathematical formula of 3d-Otsu method reasoning get recurrence formula, reduce the running time of the algorithm.The improved particle swarm optimization algorithm in this paper is used with the improved three-dimensional Otsu method for image segmentation, the original image and the image which is added noise. The results show that the method in this paper used in the noise image segmentation, can reduce the time of the segmentation and the noise of the image and keep the edge and the more detail information.