An automatic method of building extraction from high resolution satellite images

Xianwei Han; Yili Fu; Zhiqiang Qin; Kun Xing

doi:10.1117/12.832599

Translator Disclaimer

30 October 2009 An automatic method of building extraction from high resolution satellite images

Xianwei Han, Yili Fu, Zhiqiang Qin, Kun Xing

Author Affiliations +

Proceedings Volume 7495, MIPPR 2009: Automatic Target Recognition and Image Analysis; 749540 (2009) https://doi.org/10.1117/12.832599
Event: Sixth International Symposium on Multispectral Image Processing and Pattern Recognition, 2009, Yichang, China

Abstract

In this paper, we propose a novel framework of building extraction from high resolution satellite remote sensing images. This approach combines regional information with edge features, which could avoid the limitations of traditional methods. Firstly, the irrelevant regions are filtered out from large images by extracting texture features, leaving buildings distributed in the remaining urban areas. Secondly, boundaries of objects are extracted and all the contours are tracked through the chain code method we developed. Finally, the short straight lines are linked to construct contours and extract rectangular building rooftops. Experiments show that this extraction strategy is robust and effective.

Citation Download Citation

Xianwei Han, Yili Fu, Zhiqiang Qin, and Kun Xing "An automatic method of building extraction from high resolution satellite images", Proc. SPIE 7495, MIPPR 2009: Automatic Target Recognition and Image Analysis, 749540 (30 October 2009); https://doi.org/10.1117/12.832599

ACCESS THE FULL ARTICLE