基于主成分分析与粒子群优化的遥感影像变化检测  

Change detection method of remote sensing image based on PCA and PSO algorithm

作　　者：许石罗 牛瑞卿[1] 武雪玲[1] 刘超贤 

机构地区：[1]中国地质大学地球物理与空间信息学院,武汉430074

出　　处：《测绘科学》2017年第4期151-156,共6页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(41501470)

摘　　要：针对变化检测中差异影像上变化阈值选择困难的问题,该文提出了使用粒子群优化算法自动地从差异影像上选择最优变化阈值的方法。该方法首先利用主成分分析法从影像中提取包含最大信息的主分量,利用主分量构建差异影像;在此基础上,使用高斯混合模型估算差异影像的初始阈值,最后通过粒子群算法计算出最优的变化阈值。实验结果表明,该方法能够迅速地获取变化阈值,避免了人工选择时繁琐重复的尝试,且检测结果精度较高。It is difficult to determine the changing threshold from difference image. In this paper, the particle swarm optimization（PSO）algorithm was used to choose optimal change threshold from the differ- ence image. Firstly, the difference image was made by principal component analysis （PCA）method, which could extract the maximum information component from the remote sensing image. Then, PSO al- gorithm was used to obtain the optimal change threshold automatically and rapidly. Comparison result with the post-classification and CVA change detection methods indieated that the proposed method had higher aceuraey and was more effective.

关 键 词：变化检测 粒子群优化PSO 主成分分析 遥感影像 

分 类 号：P208[天文地球—地图制图学与地理信息工程]