地球信息科学学报 ›› 2015, Vol. 17 ›› Issue (4): 478-485.doi: 10.3724/SP.J.1047.2015.00478

• • 上一篇    下一篇

分形定量选择遥感影像最佳空间分辨率的方法与实验

冯桂香(), 明冬萍*()   

  1. 中国地质大学(北京)信息工程学院,北京 100083
  • 收稿日期:2014-10-16 修回日期:2014-11-15 出版日期:2015-04-10 发布日期:2015-04-10
  • 通讯作者: 明冬萍 E-mail:983738303@qq.com;mingdp@cugb.edu.cn
  • 作者简介:

    作者简介:冯桂香(1990-),女,江西人,硕士生,主要从事遥感图像处理及GIS应用研究。E-mail:983738303@qq.com

  • 基金资助:
    国家自然科学基金项目(41371347);中央高校基本科研业务费专项资金(2652013084)

Fractal Based Method on Selecting the Optimal Spatial Resolution for Remote Sensing Image

FENG Guixiang(), MING Dongping*()   

  1. School of Information Engineering, China University of Geosciences (Beijing), Beijing 100083, China
  • Received:2014-10-16 Revised:2014-11-15 Online:2015-04-10 Published:2015-04-10
  • Contact: MING Dongping E-mail:983738303@qq.com;mingdp@cugb.edu.cn
  • About author:

    *The author: SHEN Jingwei, E-mail:jingweigis@163.com

摘要:

遥感影像观测尺度是遥感信息提取研究的重要内容之一,也是遥感信息提取的焦点。以往,遥感影像尺度特征的分析大多基于地统计学,其主要体现遥感影像中的线性特征,而实质上遥感影像中既存在线性特征,又存在非线性特征。因此,在深入剖析遥感影像尺度效应及分形特征机理的基础上,本文探讨了分形理论定量选择遥感影像最佳空间分辨率(也称最佳像元观测尺度)的方法。以IKONOS全色影像的建筑用地、耕地、林地为研究对象,分别使用FBM、DBM、TPM 3种分形维数计算模型,实现了3种地物在不同空间分辨率下分形维数的计算。实验结果表明,每种地物的分形维数是随空间分辨率的增大,总体呈下降趋势,且在某些特征空间分辨率上会出现拐点。从遥感影像尺度效应分析可知,遥感影像空间格局随尺度的不同,其内部结构也不同。且随着尺度的增大,很多细节将会被忽略,影像的粗糙度也随着降低。而分形维数是目前为止描述对象自相似性和不规则度的唯一基本量化值,其直观上与物体表面的粗糙程度相吻合。因此,这些拐点对应的分形维数对地物的最佳空间分辨率的选择具有一定指示意义。通过本文研究可知,使用分形理论方法研究遥感影像最佳空间分辨率(或最佳像元观测尺度),打破以往观测尺度方法研究范畴,从不同角度去分析遥感影像观测尺度问题对GIS研究与地学应用具有一定的理论和指导意义。

关键词: 最佳空间分辨率, 像元观测尺度, 分形, 分形维数计算模型, 分形维数

Abstract:

Observation scale research is one of the important subjects of scale research in remote sensing, and it is also one of the research focuses of information extraction from remote sensing images. Analysis of the scale properties in remote sensing image is normally based on geo-statistics, which mainly highlights the linear features of the remote sensing image. However, remote sensing image generally consists of both linear and non-linear features, it is insufficient and to use analysis that based on only geo-statistics. This paper discusses a fractal based method on selecting the optimal spatial resolution for remote sensing image (also known as optimal pixel-based observation scale) by analyzing the remote sensing image scale effect and study the mechanism of fractal characteristics. Three categories of study regions, covered by building, farmland and forest respectively, were cut from IKONOS panchromatic image and used as the experimental data. Then, a series of fractal dimensions based on Fractal Brown Motion, Double Blanket Method and Triangular Prism Method respectively were calculated along with the change of spatial resolution. The statistical analyses of the experimental results demonstrate that the fractal dimensions generally show a decreasing trend with the increase of spatial resolution, and some turning points emerged at certain spatial resolutions. According to the analysis, the spatial patterns or internal structures in remote sensing image vary among different scales. And with the decreasing of spatial resolution, the roughness of image will also decrease since many details are ignored. Nevertheless, the fractal dimension is the only basic quantitative value to describe the self-similarity and irregular degree of object, and it is intuitively consistent with the roughness. Therefore, the turning points at certain spatial resolutions indicate the significance for choosing the optimal spatial resolution. The experimental results show that the fractal based method on selecting the optimal observation scale is theoretically and practically significant to geo-applications, and it extends the research categories by analyzing remote sensing observation scales from different perspectives.

Key words: optimal spatial resolution, pixel-based observation scale, fractal, fractal dimension calculation model, fractal dimension