多种分类器结合的遥感影像分类研究

Research on Remote Sensing Image Classification Combined with Multiple Classifiers

通过对遥感影像的分类方法及原理的分析,以多光谱遥感影像为依据,采集农田、菜地、居民地等三类地物样本。利用Matlab软件,分别使用最短距离分类器、贝叶斯分类器及BP神经网络分类器三种分类器将遥感影像进行分类实验。通过投票法,融合不同分类器最终的输出结果,将多种分类器的遥感影像分类进行整合,提高了影像的分类精度。

This paper mainly studies the method and basic principle of remote sensing image classification combined with multi classifiers.Based on the multi spectral remote sensing data,the samples are collected.Three kinds of ground objects,including farmland,vegetable land and residential area,are selected.Under Matlab,the experiment of remote sensing image classification is completed by using the shortest distance classifier,Bayesian classifier and BP neural network classifier.Then,by voting,the output of different classifiers is fused to realize the remote sensing image classification combined with multiple classifiers,which can greatly improve the classification accuracy of remote sensing image.