Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XL-3, 183-189, 2014
© Author(s) 2014. This work is distributed
under the Creative Commons Attribution 3.0 License.
11 Aug 2014
Morphological operation based dense houses extraction from DSM
Y. Li1, L. Zhu2, K. Tachibana2, and H. Shimamura2 1Jiangsu Provincial Key Laboratory of Geographic Information Science and Technology International Institute for Earth System Science, Nanjing University, 210000 Nanjing Jiangsu, China
2PASCO corporation Japan, 1-1-2 Higashiyama, Megulo-ku Tokyo 153-0043, Japan
Keywords: Mathematics, Transformation, Estimation, Detection, Building, Reconstruction, DSM Abstract. This paper presents a method of reshaping and extraction of markers and masks of the dense houses from the DSM based on mathematical morphology (MM). Houses in a digital surface model (DSM) are almost joined together in high-density housing areas, and most segmentation methods cannot completely separate them. We propose to label the markers of the buildings firstly and segment them into masks by watershed then. To avoid detecting more than one marker for a house or no marker at all due to its higher neighbour, the DSM is morphologically reshaped. It is carried out by a MM operation using the certain disk shape SE of the similar size to the houses. The sizes of the houses need to be estimated before reshaping. A granulometry generated by opening-by-reconstruction to the NDSM is proposed to detect the scales of the off-terrain objects. It is a histogram of the global volume of the top hats of the convex objects in the continuous scales. The obvious step change in the profile means that there are many objects of similar sizes occur at this scale. In reshaping procedure, the slices of the object are derived by morphological filtering at the detected continuous scales and reconstructed in pile as the dome. The markers are detected on the basis of the domes.
Conference paper (PDF, 736 KB)

Citation: Li, Y., Zhu, L., Tachibana, K., and Shimamura, H.: Morphological operation based dense houses extraction from DSM, Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XL-3, 183-189, doi:10.5194/isprsarchives-XL-3-183-2014, 2014.

BibTeX EndNote Reference Manager XML