基于CART集成学习的城市不透水层百分比遥感估算

基金项目: 国家自然科学基金资助项目(60472039,40523005)

详细信息

Estimating Urban Impervious Surface Percent Using Boosting as a Refinement of CART Analysis

¹ State Key Laboratory of Information Engineering in Surveying,Mapping and Remote Sensing,Wuhan University,129 Luoyu Road,Wuhan 430079,China¹ Institute of Space and Earth Information Science,Chinese University of Hong Kong,Shatin,New Territories,Hong Kong,China

Funds: 国家自然科学基金资助项目(60472039,40523005)

摘要: 利用Landsat ETM+遥感数据,提出了一种基于CART集成学习的ISP遥感亚像元估算方法,将Boosting重采样技术引入CART分析中,用于提高ISP估算的精度。实验结果表明,该方法的ISP估算性能优于传统的单一CART学习算法,从ETM+影像中估算的ISP值与真实值之间的相关系数达到0.91,平均偏差为11.16%。
- 城市不透水层 /
- 遥感影像 /
- 分类与回归树 /
- Boosting技术 /
- 集成学习
Abstract: An approach for estimating urban imperviousness surface percent(ISP) through the synergistic use of Landsat ETM+ and high-resolution imagery is presented.In the proposed approach,the ensemble leaning of CART analysis based on Boosting is utilized to construct the ISP predict model,furthermore,to obtain the subpixel ISP results at 30 m resolution.The experiment shows this approach yields ISP estimation performance compared to that of the traditional estimation method based on single CART,and its correlation coefficient of predicted versus actual ISP reaches 0.91 with an average error 11.16%.
- urban impervious surface /
- remote sensing imagery /
- classification and regression tree(CART) /
- Boosting /
- ensemble learning