资源与产业 ›› 2012, Vol. 14 ›› Issue (1): 139-143.

• 资源经济 • 上一篇    下一篇

基于Quickbird数据的土地覆盖类型分形特征研究

袁涛1,冯聪2,周伟1   

  1. 1中国地质大学 土地科学技术学院,北京 100083;2中国国土资源经济研究院,北京101149
  • 收稿日期:2011-04-09 修回日期:2011-11-09 出版日期:2012-01-15 发布日期:2012-01-15
  • 作者简介:袁涛(1981—),男,博士、讲师,主要从事土地信息系统、国土资源遥感应用研究。E-mail:yuantao@cugb.edu.cn
  • 基金资助:

    国家高技术研究发展计划(863计划)专项经费项目(2008AA1212014)

FRACTAL FEATURES OF LAND COVERAGE TYPES BASED ON QUICKBIRD DATA

YUAN Tao1, FENG Cong2, ZHOU Wei1   

  1. 1. School of Land Science and Technology, China University of Geosciences, Beijing 100083, China; 2. Chinese Institute of Land Resources and Economy, Beijing 101149, China
  • Received:2011-04-09 Revised:2011-11-09 Online:2012-01-15 Published:2012-01-15

摘要:

分形维数可以有效地描述复杂纹理特征,传统的遥感自动分类主要利用数据的波谱特征信息,很难充分有效的利用遥感影像的纹理信息。近年来,随着高分辨率遥感数据的广泛使用,如何量化纹理特征,根据不同地物的纹理信息进行地物的分类等问题引起很多学者的关注。在对比不同边缘提取算子提取效果的基础上,对4种典型土地覆盖类型的Quickbird数据进行边缘提取,并在MATLAB平台下计算影像样本盒维数,通过对比发现这些土地覆盖类型盒维数差异显著,在此基础上提出利用分形维数对遥感影像的纹理信息进行量化,从而利用影像纹理信息对不同土地覆盖类型进行有效区分。

关键词: 土地覆盖, 分型特征, 边缘提取

Abstract:

Fractal dimension can effectively describe complex textural features. The traditional automatic classification of remote sensing is mainly based on the spectrum feature information of data, which is difficult to utilize the textural information of remote sensing image throughly and effectively. With the wide application of HR remote sensing data in recent years, problems such as how to quantize textural features, how to classify ground objects according to their respective textural features are arousing many scholars’ concern. Based on the comparison of different effects of operator extraction on different edges, the study takes edge extractions on Quickbird data of four kinds of typical land cover types, and calculates dimensions of image sample boxes on the platform of Matlab. It is found of significant difference for box dimensions of these land cover classifications by comparison, thus fractal dimension can be utilized to quantize textural features of remote sensing images and by which, different land cover types can be classified effectively.

Key words: land cover, fractal feature, edge extraction

中图分类号: