Conference Proceeding

Reconstructing building interiors from images

University of Washington, Seattle, USA
Proceedings / IEEE International Conference on Computer Vision. IEEE International Conference on Computer Vision 11/2009; DOI:10.1109/ICCV.2009.5459145 pp.80 - 87 In proceeding of: Computer Vision, 2009 IEEE 12th International Conference on
Source: IEEE Xplore

ABSTRACT This paper proposes a fully automated 3D reconstruction and visualization system for architectural scenes (interiors and exteriors). The reconstruction of indoor environments from photographs is particularly challenging due to texture-poor planar surfaces such as uniformly-painted walls. Our system first uses structure-from-motion, multi-view stereo, and a stereo algorithm specifically designed for Manhattan-world scenes (scenes consisting predominantly of piece-wise planar surfaces with dominant directions) to calibrate the cameras and to recover initial 3D geometry in the form of oriented points and depth maps. Next, the initial geometry is fused into a 3D model with a novel depth-map integration algorithm that, again, makes use of Manhattan-world assumptions and produces simplified 3D models. Finally, the system enables the exploration of reconstructed environments with an interactive, image-based 3D viewer. We demonstrate results on several challenging datasets, including a 3D reconstruction and image-based walk-through of an entire floor of a house, the first result of this kind from an automated computer vision system.

0 0
 · 
0 Bookmarks
 · 
30 Views
  • Source
    Conference Proceeding: A Cost-Efficient 3D Sensing System for Autonomous Mobile Robots
    Marco A Gutiérrez, E Martinena, A Sánchez, Rosario G Rodríguez, P Núñez
    [show abstract] [hide abstract]
    ABSTRACT: This paper describes a mechanism for building an inexpensive and, at the same time, accurate system for 3D scanning on Autonomous Mobile Robots. Our system allows us to obtain 3D points from the robot environment along with its associated color. This data can be later processed using different techniques in order to obtain information from surrounding objects useful for tasks such as navigation or localization. Information is obtained at a rate of 50 ms per line of scan (700 points per line). In order to use the sensor as part of an active perception system, resolution is made to be directly dependent on the scanning speed and robots are able to adjust the related parameters accordingly to their needs. Our approach uses a regular commercial 2D Laser Range Finder (LRF), a step motor and a camera, all this controlled by an embedded circuit which makes the system apt for being built in any regular Autonomous Mobile Robot. Finally, to test our system, two different real applications have been used. First a 3D Map reconstruction is done using several point clouds matched by the ICP algorithm and our odometry. Then, we make a novelty detection and 3D shape retrieval using the Gaussian Mixture Model and Superquadrics.
    XII WORKSHOP OF PHYSICAL AGENTS 2011; 09/2011
  • Source
    Conference Proceeding: Automatic reconstruction of textured 3D models
    B. Pitzer, S. Kammel, C. DuHadway, J. Becker
    [show abstract] [hide abstract]
    ABSTRACT: This paper describes a system for automatic mapping and generation of textured 3D models of indoor environments without user interaction. Our data acquisition system is based on a Segway RMP platform which allows us to automatically acquire large amounts of textured 3D scans in a short amount of time. The first data processing step is registration and mapping. We propose a probabilistic, non-rigid registration method that incorporates statistical sensor models and surface prior distributions to optimize alignment and the reconstructed surface at the same time. Second, in order to fuse multiple scans and to reconstruct a consistent 3D surface representation, we incorporate a volumetric surface reconstruction method based on a oriented point. For the final step of texture reconstruction, we present a novel method to automatically generate blended textures from multiple images and multiple scans which are mapped onto the 3D model for photo-realistic visualization. We conclude our report with results from a large-scale, real-world experiment. The most significant contribution of this research is a functional system that covers all steps required to automatically reconstruct textured 3D models of large indoor environments.
    Robotics and Automation (ICRA), 2010 IEEE International Conference on; 06/2010
  • Source
    Conference Proceeding: Dense multi-planar scene estimation from a sparse set of images.
    Alberto Argiles, Javier Civera, Luis Montesano
    2011 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2011, San Francisco, CA, USA, September 25-30, 2011; 01/2011

Show more

Data provided are for informational purposes only. Although carefully collected, accuracy cannot be guaranteed. The impact factor represents a rough estimation of the journal's impact factor and does not reflect the actual current impact factor. Publisher conditions are provided by RoMEO. Differing provisions from the publisher's actual policy or licence agreement may be applicable.

Keywords

3D model
 
3D reconstruction
 
architectural scenes
 
automated 3D reconstruction
 
automated computer vision system
 
challenging datasets
 
depth maps
 
dominant directions
 
entire floor
 
first result
 
image-based 3D viewer
 
image-based walk-through
 
initial 3D geometry
 
Manhattan-world assumptions
 
Manhattan-world scenes
 
novel depth-map integration algorithm
 
piece-wise planar surfaces
 
simplified 3D models
 
texture-poor planar surfaces
 
uniformly-painted walls