地球信息科学学报

地球信息科学学报 ›› 2010, Vol. 12 ›› Issue (6): 855-862.

• 本期要文(可全文下载) • 上一篇    下一篇

北京一号小卫星数据的城市景观格局监测分析——以徐州市城区为例

杜培军, 单丹丹, 夏俊士, 刘培   

  1. 国土环境与灾害监测国家测绘局重点实验室(中国矿业大学), 徐州 221116
  • 收稿日期:2009-11-24 修回日期:2010-10-25 出版日期:2010-12-25 发布日期:2010-12-25
  • 作者简介:杜培军,男,山西省五台县人,工学博士,教授,博士生导师,从事遥感信息工程、地理信息科学与技术领域等方 面的研究。E-mail:dupjrs@cumt.edu.cn
  • 基金资助:

    国家“863”高技术研究发展计划(2007AA12Z162);教育部新世纪优秀人才支持计划资助项目(NCET-06-0476);北京宇视蓝图信息技术有限公司北京一号小卫星应用开放基金资助项目(2007-09)

Monitoring and Analyzing Urban Landscape Pattern Change Using Beijing-1 Small Satellite Remote Sensing Data:Taking the Urban Area of Xuzhou City as an Example

DU Peijun, SHAN Dandan, XIA Junshi, LIU Pei   

  1. Key Laboratory for Land Environment and Disaster Monitoring of State Bureau of Surveying and Mapping,China University of Mining and Technology,Xuzhou 221116,China
  • Received:2009-11-24 Revised:2010-10-25 Online:2010-12-25 Published:2010-12-25

PDF (PC)

赞

摘要： 为了推进北京一号小卫星遥感数据在城市景观生态研究中的应用,针对其全色影像4m分辨率、多光谱影像32m分辨率的特点,试验分析北京一号小卫星影像在城市景观格局变化中的应用效果和特点。通过对不同分类器的比较,选择支持向量机分类器对多光谱影像、全色影像、全色与多光谱融合影像三个数据集进行景观组分分类,结果表明,全色与多光谱融合影像的分类精度最高。利用多时相、多光谱遥感数据统计分析了城市景观组分与格局变化,表明32m空间分辨率的多光谱影像可以用于城市景观格局变化和土地覆盖变化分析。本文全面试验和分析评价了北京一号多分辨率数据在城市景观格局研究中的应用效果。通过对同一年份全色影像和多光谱影像计算的景观格局指标的分析表明,全色数据能更有效地描述景观详细信息,多光谱数据可展现城市景观的整体格局,而融合后对景观格局分析能够获得优于单一数据的效果。试验和分析表明,北京一号小卫星4m全色高分辨率影像和32m多光谱数据的波段组合,能从不同尺度揭示城市景观格局和变化过程。

关键词: 北京一号小卫星, 景观格局, 支持向量机, 融合, 分类

Abstract: In order to push the applications of Beijing-1 small satellite remote sensing data to urban landscape ecology studies,the panchromatic high resolution data,medium resolution multi-spectral data and fused data of panchromatic and multispectral images of Beijing-1 were used to analyze urban landscape pattern from class level to landscape level.By comparing different classifiers,support vector machine(SVM) was selected to classify the land cover and landscape component types,and spectral and textural features were integrated to improve classification accuracy.After a comparison to some fusion algorithms including IHS,Brovey,PCA and Gram-Shmidt algorithm,we chosen IHS algorithm to fuse the panchromatic and multispectral images.The classification results of multispectral images on two dates were employed for landscape change analysis at first,and the results showed that the landscape from 2007 to 2008 had obvious changes,which were also evidenced by the statistics of landscape metrics from landscape level to class level.By comparing the landscape pattern indices derived from panchromatic and multispectral images,we can find that the overall properties keep similar on two data sets,but 4m panchromatic image was good for revealing more detailed features,and the 32m multispectral images were effective to show the overall trends and patterns.The fusion image of panchromatic and multispectral images outperform any individual data sets in analyzing urban landscape pattern owing to the combination of high spatial resolution and multispectral capacity.The results show that Beijing-1 small satellite remote sensing data are effective to monitor and analyze urban landscape pattern,especially,the band allocation of 4m panchromatic image and 32m multispectral images can describe and reveal the landscape pattern and process on different levels.

Key words: Beijing-1 small satellite, landscape pattern, support vector machine, fusion, classification