Reconstruction & Analysis

Technical Papers

Reconstruction & Analysis

Wednesday, 12 August 10:45 AM - 12:15 PM | Los Angeles Convention Center, Room 153A-C Session Chair: Hao Li, University of Southern California


RAPter: Rebuilding Man-Made Scenes With Regular Arrangements of Planes

Gathering large point clouds is now easy, but processing remains a challenge. Raw scans with missing data, noise, and varying sampling density are particularly a problem. RAPter reconstructs man-made scenes as regular arrangements of planes, selecting both local approximations along with their global inter-relations, while robustly avoiding over-regularization.

Aron Monszpart
University College London

Nicolas Mellado
Centre national de la recherche scientifique, University College London, Université Paul Sabatier

Gabriel Brostow
University College London

Niloy J. Mitra
University College London

LOD Generation for Urban Scenes

Starting from a raw surface mesh, this approach generates four compact levels of detail that are meaningful, abstracted, and enriched with urban semantics.

Florent Lafarge
INRIA

Yannick Verdie
École polytechnique fédérale de Lausanne

Pierre Alliez
INRIA

Coupled Segmentation and Similarity Detection for Architectural Models

In this paper, a novel algorithm for simultaneous segmentation and labeling of unstructured architectural models is formulated as a set-cover problem and solved using an extension of constrained combinatorial optimization.

Ilke Demir
Purdue University

Daniel Aliaga
Purdue University

Bedrich Benes
Purdue University

Shape Segmentation by Approximate Convexity Analysis

A shape-segmentation method for complete and incomplete shapes and point clouds based on a characterization of the expected geometry of shape parts.

Oliver van Kaick
Carleton University

Noa Fish
Tel Aviv University

Yanir Kleiman
Tel Aviv University

Shmuel Asafi
Tel Aviv University

Daniel Cohen-Or
Tel Aviv University