计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (15): 12-14.

• 博士论坛 • 上一篇    下一篇

基于蚁群优化分类规则挖掘的遥感图像分类研究

戴 芹,刘建波   

  1. 中国科学院 中国遥感卫星地面站重点实验室,北京 100086
  • 收稿日期:2007-12-13 修回日期:2008-03-04 出版日期:2008-05-21 发布日期:2008-05-21
  • 通讯作者: 戴 芹

Research on remote sensing image classification using Ant Colony Optimization(ACO) based classification rule mining algorithm

DAI Qin,LIU Jian-bo   

  1. The Key Laboratory of China Remote Sensing Satellite Ground Station,CAS,Beijing 100086,China
  • Received:2007-12-13 Revised:2008-03-04 Online:2008-05-21 Published:2008-05-21
  • Contact: DAI Qin

摘要: 蚁群优化算法作为群智能理论的主要算法之一,已经成功应用在众多研究领域的优化问题上,但是在遥感数据处理领域还是一个新的研究课题。蚁群优化具有自组织、合作、通信等智能化优点,对数据无需统计分布参数的先验知识,因此在遥感数据处理领域具有很大的潜在优势。介绍了将蚁群优化分类规则挖掘算法应用到遥感图像分类研究领域的理论与算法流程。并采用北京地区的CBERS遥感数据作为实验数据,通过蚁群优化算法构造分类规则,对选择的遥感数据进行了分类实验,并和最大似然分类方法进行对比,实验结果表明,蚁群优化分类规则挖掘算法为遥感图像的分类提供了一种新方法。

关键词: 蚁群优化算法, 分类规则挖掘, 遥感图像分类

Abstract: Ant Colony Optimization(ACO) as one of the main algorithms in swarm intelligence has been applied successfully to optimization problems in many research areas,but it is still a new research topic in remote sensing data processing.The ACO algorithm has self-organization,cooperation,communication,and other intelligent merits,and without prior knowledge of statistical distribution parameters,so it has many potential advantages in the remote sensing data processing research area.This paper introduces the theory and algorithm process of the ACO for mining classification rules applied to remote sensing image classification.Selecting CBERS image located in Beijing area as experimental data,this paper use the ACO based rule mining algorithm for constructing the classification rules,then use these rules to classify the example data,and compared the classification result with Maximum Likelihood Classification(MLC) method,the experiment results show that ACO based classification rule mining algorithm has provided a new method for remote sensing image classification.

Key words: Ant Colony Optimization(ACO) algorithm, classification rule mining, remote sensing image classification