跳到主要导航 跳到搜索 跳到主要内容

Urban Building Density Estimation From High-Resolution Imagery Using Multiple Features and Support Vector Regression

科研成果: 期刊稿件 › 论文 › 同行评审

56 引用 (Scopus)

摘要

Building density is considered as an environmental parameter of great significance for urban management, and it can also help us to understand how cities function. However, in developing countries, building density information is often lacking or incomplete in a large number of cities. To address this problem, this paper presents a method for the quantitative estimation of building density using a support vector regression model to establish the relationship between building density and the features extracted from the image, including spectral, morphological, and textural features. The importance and relevance of the different features is investigated by a feature selection method named recursive feature elimination. Tests performed on three representative megacities in China confirm that the proposed method achieves satisfactory results for building density estimation in terms of the low root-mean-square errors (similar to 0.05) and visual effect. It is also found that the combination of feature sets outperforms the single feature set, indicating the need for simultaneous consideration of all the categories of features. The knowledge of the most informative features provides an insight into selecting the most effective parameters for building density mapping. The investigation of the transferability between geographical regions confirms that the self-training achieves the better performance compared to the transfer training. The results with regard to the regression accuracy, feature selection, and the comparison with other methods indicate the effectiveness of the multifeature approach when applied to building density estimation.
源语言英语
页(从-至)3265-3280
页数16
期刊IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
卷10
期7
DOI
出版状态已出版 - 7月 2017

基金资助

This work was supported in part by the National Key Research and Development Program of China under Grant 2016YFB0501403, in part by the China Science Fund for Excellent Young Scholars under Grant 41522110, and in part by the Foundation for the Author of National Excellent Doctoral Dissertation of PR China under Grant 201348. (Corresponding authors: Xin Huang and Dawei Wen.)

关键词

  • Building density
  • feature selection
  • mathematical morphology
  • remote sensing
  • support vector regression (SVR)
  • texture analysis

收录

  • EI
  • Scopus
  • SCIE

学术指纹

探究 'Urban Building Density Estimation From High-Resolution Imagery Using Multiple Features and Support Vector Regression' 的科研主题。它们共同构成独一无二的学术指纹。

引用此