Application Of Adaptive Particle Swarm Optimization Algorithm In Ceramic Image Segmentation

Visual image is an important source for human to obtain information.With the continuous enhancement of computer processing ability,the extensive application of image processing has the hardware foundation that can be realized.Image processing technology has been widely used in industry,agriculture and medicine.Among them,image segmentation plays a very important role in image analysis,image recognition,image detection and other aspects,and has always been a hot topic for many scholars to study.As a kind of bionic intelligence algorithm,Particle Swarm Optimization（PSO）has the advantages of strong search ability,easy to adjust and relatively independent related parameters in the algorithm.It has been widely used in pattern recognition,fault diagnosis,machine learning,resource scheduling and other applications.However,particle swarm optimization algorithm is applied to the optimization process,there are often late slow convergent speed and easy to fall into local most superior defects,this paper proposes an adaptive particle swarm optimization algorithm with extremal disturbance（DAPSO）,and combines DAPSO and most between-cluster variance method applied to the optimization problem and ceramic image segmentation process,better simulation results are obtained.The main research contents of this paper include:1.This paper introduces the basic principle of particle swarm optimization algorithm,introduces the method of parameter adaptive change of particle swarm optimization algorithm,and analyzes the influence of parameter setting in particle swarm optimization algorithm,and proposes a new inertial weight adaptive particle swarm optimization algorithm（APSO）.The algorithm takes the fitness change of particles between two adjacent iterations as the criterion and adjusts the value of inertia weight in the next iteration by integrating the flight status of particles.When the particle population falls into the local optimal region,the individual extreme value and the global extreme value are disturbed simultaneously by cauchy perturbation method,taking the population diversity and the number of evolutionary stagnation steps as the judging conditions of the extreme perturbation.In order to protect some good global optimal structure that particles may have when they fall into local optimal,only one dimension is randomly selected for perturbation,and each dimension is selected with the same probability.Experiments on standard function set show that DAPSO algorithm can accelerate the convergence rate of particle swarm and prevent particles from falling into local optimum to some extent.2.The maximum inter-class variance method in image segmentation is introduced.The maximum inter-class variance method needs to traverse every gray level in image segmentation,and its real-time performance is poor,and the time complexity will increase with the increase of gray level.In this paper,DAPSO is combined with the maximum inter-class variance method for the segmentation of ceramic images.Experiments show that the proposed hybrid algorithm not only saves the time of image segmentation,but also can better segment the texture details of ceramic images with complex textures.