Boosting和Bagging算法的高分辨率遥感影像分类探讨

多分类器集成能够有效地提高遥感分类精度、降低结果中的不确定性,基于样本操作的Boosting和Bagging算法是多分类器系统常用的两种算法。针对高分辨率卫星遥感分类的需求,以Qu ickb ird数据为例,分别以BP神经网络、RBF神经网络和决策树为基分类器,对Boosting和Bagging算法的应用效果进行了实验和分析评价,结果表明Boosting算法和Bagging算法能够用于高分辨率遥感影像分类,具有较好的分类性能。

High resolution remote sensing image classificatoin based on Boosting and Bagging algorithms

As an advanced direction in pattern recognition,the applications of multiple classifier system or multiple classifier integration in remote sensing will further improve the reliability and accuracy of classification and reduce uncertainty in results.Boosting and Bagging,as two popular classifier ensemble algorithms based on sample manipulation,are used to high resolution remote sensing image classification in this paper.By using QuickBird image as the data souce and BPNN,RBFNN and decision tree as base classifiers,the performance of Boosting and Bagging to high resolution remote sensing image classification is analyzed.The results show that Boosting and Bagging are effective to high resolution remote sensing classification.