EG2013
https://diglib.eg.org:443/handle/10.2312/298
Girona, Spain2024-03-29T00:42:53ZGuided Capturing of Multi-view Stereo Datasets
https://diglib.eg.org:443/handle/10.2312/conf.EG2013.short.093-096
Guided Capturing of Multi-view Stereo Datasets
Langguth, Fabian; Goesele, Michael
M.- A. Otaduy and O. Sorkine
We present an application for mobile devices, that allows any user, even without background in computer vision, to capture a complete set of images, that is suitable for a multi-view stereo reconstruction. Compared to related tasks, such as panorama capture, this setting is much harder, as the camera needs to move unrestricted in 3D space. Our system uses structure from motion to register captured images and generates a sparse reconstruction of the scene. The dataset is built in an incremental procedure, where the next best view is computed with a novel view planning strategy, that aims for a good coverage of the scene. The user is then guided towards the new view, and the image is captured automatically at the right position. The next iteration starts after the reconstruction has been updated. The quality of the resulting dataset is on par with datasets captured by an expert user.
2013-01-01T00:00:00ZAutomatic Modeling of Planar-Hinged Buildings
https://diglib.eg.org:443/handle/10.2312/conf.EG2013.short.089-092
Automatic Modeling of Planar-Hinged Buildings
Garcia-Dorado, Ignacio; Aliaga, Daniel G.
M.- A. Otaduy and O. Sorkine
We present a framework to automatically model and reconstruct buildings in a dense urban area. Our method is robust to noise and recovers planar features and sharp edges, producing a water-tight triangulation suitable for texture mapping and interactive rendering. Building and architectural priors, such as the Manhattan world and Atlanta world assumptions, have been used to improve the quality of reconstructions. We extend the framework to include buildings consisting of arbitrary planar faces interconnected by hinges. Given millions of initial 3D points and normals (i.e., via an image-based reconstruction), we estimate the location and properties of the building model hinges and planar segments. Then, starting with a closed Poisson triangulation, we use an energy-based metric to iteratively refine the initial model so as to attempt to recover the planar-hinged model and maintain building details where possible. Our results include automatically reconstructing a variety of buildings spanning a large and dense urban area, comparisons, and analysis of our method. The end product is an automatic method to produce watertight models that are very suitable for 3D city modeling and computer graphics applications.
2013-01-01T00:00:00ZGPU Roof Grammars
https://diglib.eg.org:443/handle/10.2312/conf.EG2013.short.085-088
GPU Roof Grammars
Buron, Cyprien; Marvie, Jean-Eudes; Gautron, Pascal
M.- A. Otaduy and O. Sorkine
We extend GPU shape grammars [MBG12] to model highly detailed roofs. Starting from a consistent roof structure such as a straight skeleton computed from the building footprints, we decompose this information into local roof parameters per input segments compliant with GPU shape grammars. We also introduce Join and Project rules for a consistent description of roofs using grammars, bringing the massive parallelism of GPU shape grammars to the benefit of coherent generation of global structures.
2013-01-01T00:00:00ZSifted Disks
https://diglib.eg.org:443/handle/10.1111/v32i2pp509-518
Sifted Disks
Ebeida, Mohamed S.; Mahmoud, Ahmed H.; Awad, Muhammad A.; Mohammed, Mohammed A.; Mitchell, Scott A.; Rand, Alexander; Owens, John D.
I. Navazo, P. Poulin
We introduce the Sifted Disk technique for locally resampling a point cloud in order to reduce the number of points. Two neighboring points are removed and we attempt to find a single random point that is sufficient to replace them both. The resampling respects the original sizing function; In that sense it is not a coarsening. The angle and edge length guarantees of a Delaunay triangulation of the points are preserved. The sifted point cloud is still suitable for texture synthesis because the Fourier spectrum is largely unchanged. We provide an efficient algorithm, and demonstrate that sifting uniform Maximal Poisson-disk Sampling (MPS) and Delaunay Refinement (DR) points reduces the number of points by about 25 percent, and achieves a density about 1/3 more than the theoretical minimum. We show two-dimensional stippling and meshing applications to demonstrate the significance of the concept.
2013-01-01T00:00:00Z