基于猫群算法的图像分类研究  

Research of image classification based on cat swarm optimization

作　　者：王光彪[1] 杨淑莹[1] 冯帆[1] 王博凯[1] 贾紫娟[1] 朱光 

机构地区：[1]天津理工大学计算机与通信工程学院,天津300384 [2]山东省广播电影电视局,山东273300

出　　处：《天津理工大学学报》2011年第5期35-39,共5页Journal of Tianjin University of Technology

基　　金：天津市高等学校科技发展基金(20071308)

摘　　要：针对传统进化算法在图像分类中存在的收敛速度慢,易陷入局部最优等问题,本文提出用猫群算法求解图像分类问题,将求解组合优化问题转化为猫群的位置寻优过程,并分析了猫群算法及其两种行为模式下的算法模型.讨论了不同模式下猫群的速度、位置更新公式,并阐述了利用该算法求解图像分类问题的具体步骤.通过实验,验证了猫群算法在图像分类中的准确性和有效性.There are much more problems such as slow convergence rate and easily fall into premature convergence in image classification based on conventional evolutionary algorithms. So this paper proposes a novel approach, Cat Swarm Optimization （CSO） , to overcome those problems. We transform our problems from combinatorial optimization to the process of cat＇s location optimization, and analysis Cat Swarm Optimization and its two models of algorithm under cat＇s behavior. We discuss the updating formula of cat＇s speed and location in different modes and elaborate the concrete steps of image classification by using this approach. Experiments show that this approach for image classification is correct and effective.

关 键 词：猫群算法 群体智能 图像分类 组合优化 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]