测绘通报 ›› 2017, Vol. 0 ›› Issue (11): 32-36.doi: 10.13474/j.cnki.11-2246.2017.0343

• 学术研究 • 上一篇    下一篇

利用多分类器自适应级联模型的高分二号影像分类

王耀1,2, 杨化超1, 王光辉1,2, 黄杰1,2, 王更1,2, 刘笑1,2   

  1. 1. 中国矿业大学, 江苏 徐州 221116;
    2. 国家测绘地理信息局卫星测绘应用中心, 北京 100830
  • 收稿日期:2017-04-04 修回日期:2017-06-15 出版日期:2017-11-25 发布日期:2017-12-07
  • 作者简介:王耀(1992-),男,硕士,研究方向为遥感信息提取。E-mail:xz_wangyao@163.com
  • 基金资助:
    国家自然科学基金(41371438)

Classification of GF-2 Image Used by Multiple Classifiers Self-adaption Cascade Model

WANG Yao1,2, YANG Huachao1, WANG Guanghui1,2, HUANG Jie1,2, WANG Geng1,2, LIU Xiao1,2   

  1. 1. China University of Mining and Technology, Xuzhou 221116, China;
    2. Satellite Surveying and Mapping Application Center, NASG, Beijing 100830, China
  • Received:2017-04-04 Revised:2017-06-15 Online:2017-11-25 Published:2017-12-07

摘要: 针对传统单一分类器分类效果不够理想,存在各自的不足,以及已有的多分类器级联模型不能根据待识别样本特征进行动态调整优化等问题,提出了一种基于多分类器自适应级联模型的遥感影像分类方法。该模型选取各类别最优分类器进行级联组合,以待识别样本在整体性能最优分类器的表现对类别最优分类器作出自适应调整,对高分辨率影像分割后的像斑对象输出类别信息。以杭州区域高分二号遥感影像进行分类试验,结果表明,本文方法相比于单一分类器及已有的级联模型具有更高的分类精度。

关键词: 多分类器, 影像分类, 自适应模型, 高分二号

Abstract: Aiming at the shortcomings of traditional single classifier and the lacking that the existing multiple classifiers cannot adjust itself dynamically according to the characteristics of unknown sample, the classification method of high resolution remote sensing image based on the self-adaption cascade model of multiple classifiers is proposed. In this model, the optimal classifier and the whole optimal classifier are selected to dynamically assemble, making an self-adaption adjustment by performance of the sample which will be classified in this classifier, and the model can produce output category information. By GF-2 image on Hangzhou area classification test, the results show that the multiple classifiers self-adaption cascade model method has a higher classification accuracy compared to a single classifier.

Key words: multiple classifiers, image classification, self-adaption model, GF-2

中图分类号: