基于人工蜂群算法的多角度遥感影像分类  

Multi-angle remote sensing image classification based on artificial bee colony algorithm

作　　者：杨雪峰[1] 叶茂[1] 毛东雷[1] YANG Xuefeng;YE Mao;MAO Donglei(College of Geography Science and Tourism,Xinjiang Normal University,Urumqi 830054,China)

机构地区：[1]新疆师范大学地理科学与旅游学院

出　　处：《国土资源遥感》2018年第3期48-54,共7页Remote Sensing for Land & Resources

基　　金：国家自然科学基金项目"塔里木河胡杨径向生长与水分关系研究"(编号:41461045)资助

摘　　要：人工蜂群(artificial bee colony,ABC)算法在最优化领域有广泛的应用,但在遥感影像分类应用中研究较少。通过使用ABC算法构建基于规则的分类系统,对塔里木河中下游多角度遥感观测数据构建的多维数据集进行分类,生成决策规则集;并与最大似然分类法(maximum likelihood classification,MLC)、C4. 5决策树法和支持向量机(support vector machine,SVM)分类结果进行比较。结果表明,ABC算法总体分类精度高于MLC和C4. 5,但低于SVM。通过对规则中分类属性的频数分析,证明使用ABC算法可有效发现多角度数据观测结果与不同土地覆被类型之间的关系。Artificial bee colony(ABC)algorithm is widely used in optimization field,but the study of the applications of the remote sensing image classification is inadequate.Through the use of ABC algorithm,the classification system was constructed on the basis of rules.The multi-dimensional data sets consisting of the multi-angle remote sensing observation data originating from the middle and lower reaches of Tarim River were investigated so as to generate the decision rules.A comparison with the classification results of the maximum likelihood method(MLC),C4.5 decision tree and support vector machine(SVM)shows that classification accuracy of ABC is higher than that of MLC and C4.5 overall,but lower than that of SVM.At the same time,through the frequency analysis of the classification attributes in the rules,it is proved that ABC can effectively discover the relationship between the results of the multi-angle data observation and different land cover types.

关 键 词：人工蜂群(ABC)算法 多角度遥感 土地覆被 塔里木河中下游 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]