Technical Papers

Urban Reconstruction & Explanation

Wednesday, 28 July | 2:00 PM - 3:30 PM | Room 403 AB
Session Chair: Brian Curless, University of Washington
SmartBoxes for Interactive Urban Reconstruction

An interactive tool for reconstructing architectural models from 3D point clouds of urban scenes. The box primitives are smart in the sense that their locations, sizes, and interval are automatically adjusted on the fly to fit to the data while respecting contextual relations with nearby similar blocks.

Liangliang Nan
Shenzhen Institute of Advanced Technology

Andrei Sharf
Shenzhen Institute of Advanced Technology

Richard Zhang
Simon Fraser University

Baoquan Chen
Shenzhen Institute of Advanced Technology

Daniel Cohen-Or
Tel-Aviv University

Non-Local Scan Consolidation for 3D Urban Scenes

Buildings often exhibit large-scale repetitions and self-similarities. The key observation in this paper is that the same geometry, scanned over multiple recurring instances, allows application of a simple and effective non-local filtering. The geometry multiplicity is fused together and projected into a base-geometry defined by clustering corresponding surfaces.

Qian Zheng
Shenzhen Institute of Advanced Technology

Andrei Sharf
Shenzhen Institute of Advanced Technology

Guo Wei
Shenzhen Institute of Advanced Technology

Yangyan Li
Shenzhen Institute of Advanced Technology

Niloy Mitra
Indian Institute of Technology Delhi

Daniel Cohen-Or
Tel-Aviv University

Baoquan Chen
Shenzhen Institute of Advanced Technology

Ambient Point Clouds for View Interpolation

View-interpolation algorithms often produce visual artifacts in regions where the 3D scene geometry is erroneous, uncertain, or incomplete. This paper introduces ambient point clouds constructed from uncertain colored-depth pixels, which help reduce these artifacts while providing non-photorealistic background coloring and enhancing the sense of 3D motion.

Michael Goesele
Technische Universität Darmstadt

Jens Ackermann
Technische Universität Darmstadt

Simon Fuhrmann
Technische Universität Darmstadt

Carsten Haubold
Technische Universität Darmstadt

Ronny Klowsky
Technische Universität Darmstadt

Drew Steedly
Microsoft Corporation

Richard Szeliski
Microsoft Research

Street Slide: Browsing Street-Level Imagery

Street Slide is a novel browsing interface for street-level imagery that combines the best aspects of the immersive nature of bubbles with the overview provided by multi-perspective strip panoramas.

Johannes Kopf
Microsoft Research Redmond

Billy Chen
Microsoft Corporation

Richard Szeliski
Microsoft Research

Michael F. Cohen
Microsoft Research

In the News
SIGGRAPH 2010 Video