3D Building Modelling with Digital Map, Lidar Data and Video Image Sequences

Abstract

Three‐dimensional (3D) reconstruction and texture mapping of buildings or other man‐made objects are key aspects for 3D city landscapes. An effective coarse‐to‐fine approach for 3D building model generation and texture mapping based on digital photogrammetric techniques is proposed. Three video image sequences, two oblique views of building walls and one vertical view of building roofs, acquired by a digital video camera mounted on a helicopter, are used as input images. Lidar data and a coarse two‐dimensional (2D) digital vector map used for car navigation are also used as information sources. Automatic aerial triangulation (AAT) suitable for a high overlap image sequence is used to give initial values of camera parameters of each image. To obtain accurate image lines, the correspondence between outlines of the building and their line features in the image sequences is determined with a coarse‐to‐fine strategy. A hybrid point/line bundle adjustment is used to ensure the stability and accuracy of reconstruction. Reconstructed buildings with fine textures superimposed on a digital elevation model (DEM) and ortho‐image are realistically visualised. Experimental results show that the proposed approach of 3D city model generation has a promising future in many applications.

Assessing the Accuracy of Georeferenced Point Clouds Produced via Multi-View Stereopsis from Unmanned Aerial Vehicle (UAV) Imagery

“…When all disk points are extracted, the reference points for the point clusters (an example of which is shown in Figure 7(a)) need to be determined to identify the centre coordinate for each disk. An alternative approach may be to use least squares template matching [55][56][57] or ellipse fitting [58] to determine corresponding GCP locations in multiple images and then compute 3D centre point coordinates in the arbitrary coordinate system based points in the cloud (found using cluster extraction) and their matched feature descriptor vectors (containing corresponding image coordinates). This has not been attempted here and is being considered for future studies.…”

Aerial Reconstructions via Probabilistic Data Fusion

“…LiDAR has been exploited extensively in aerial reconstructions [10,15,19,22,23,24,25,38,39,40]. However, in sharp contrast to the proposed method, these methods assume that LiDAR provides a noise-free and accurate geometry, relegating images solely as a source of texture information and neglecting image-based geometric information.…”

An automatic mosaicking method for building facade texture mapping using a monocular close-range image sequence

“…Furthermore, an image sequence can be extracted from video to reconstruct a 3D geometric and texture model of a building efficiently. Zhang et al (2005) reported an approach for rapidly creating a textured 3D model of a building by combining helicopter-based video, Lidar data, and a 2D vector map. However, there were some limitations due to the helicopter-based video used.…”