Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XL-7/W4, 109-112, 2015
http://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XL-7-W4/109/2015/
doi:10.5194/isprsarchives-XL-7-W4-109-2015
© Author(s) 2015. This work is distributed
under the Creative Commons Attribution 3.0 License.
 
26 Jun 2015
Segmentation and Reconstruction of Buildings with Aerial Oblique Photography Point Clouds
P. Liu1, Y. C. Li1,2, W. Hu1, and X. B. Ding1,2 1China TopRS technology Co.Ltd., Beijing, China
2Chinese Academy of Survey & Mapping, Beijing, China
Keywords: Oblique photography, stereo matching, point clouds segmentation, building reconstruction Abstract. Oblique photography technology as an excellent method for 3-D city model construction has brought itself to large-scale recognition and undeniable high social status. Tilt and vertical images with the high overlaps and different visual angles can produce a large number of dense matching point clouds data with spectral information. This paper presents a method of buildings reconstruction with stereo matching dense point clouds from aerial oblique images, which includes segmentation of buildings and reconstruction of building roofs. We summarize the characteristics of stereo matching point clouds from aerial oblique images and outline the problems with existing methods. Then we present the method for segmentation of building roofs, which based on colors and geometrical derivatives such as normal and curvature. Finally, a building reconstruction approach is developed based on the geometrical relationship. The experiment and analysis show that the methods are effective on building reconstruction with stereo matching point clouds from aerial oblique images.
Conference paper (PDF, 1051 KB)


Citation: Liu, P., Li, Y. C., Hu, W., and Ding, X. B.: Segmentation and Reconstruction of Buildings with Aerial Oblique Photography Point Clouds, Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XL-7/W4, 109-112, doi:10.5194/isprsarchives-XL-7-W4-109-2015, 2015.

BibTeX EndNote Reference Manager XML