The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLI-B3
https://doi.org/10.5194/isprs-archives-XLI-B3-625-2016
https://doi.org/10.5194/isprs-archives-XLI-B3-625-2016
09 Jun 2016
 | 09 Jun 2016

AUTOMATIC BUILDING EXTRACTION AND ROOF RECONSTRUCTION IN 3K IMAGERY BASED ON LINE SEGMENTS

A. Köhn, J. Tian, and F. Kurz

Keywords: Building detection, line segment detection, DSM, morphological filtering, RANSAC, plane fitting, vegetation identification

Abstract. We propose an image processing workflow to extract rectangular building footprints using georeferenced stereo-imagery and a derivative digital surface model (DSM) product. The approach applies a line segment detection procedure to the imagery and subsequently verifies identified line segments individually to create a footprint on the basis of the DSM. The footprint is further optimized by morphological filtering. Towards the realization of 3D models, we decompose the produced footprint and generate a 3D point cloud from DSM height information. By utilizing the robust RANSAC plane fitting algorithm, the roof structure can be correctly reconstructed. In an experimental part, the proposed approach has been performed on 3K aerial imagery.