Abstract

This work deals with the problem of automatically choosing the correct exposure (or integration) time for time-of-flight depth image capturing. We apply methods known from high dynamic range imaging to combine depth images taken with differing integration times in order to produce high quality depth maps. We evaluate the quality of these depth maps by comparing the performance in reconstruction of planar textured patches and in the 3D reconstruction of an indoor scene. Our solution is fast enough to capture the images at interactive frame rates and also flexible to deal with any amount of exposures.

Conference article

Exposure Fusion for Time-Of-Flight Imaging (local copy, publisher page)

BibTeX

@inproceedings{HA2011,
author = {Uwe Hahne and Marc Alexa},
title = {{Exposure Fusion for Time-Of-Flight Imaging}},
pages = {1887-1894},
URL = {http://diglib.eg.org/EG/CGF/volume30/issue7/v30i7pp1887-1894.pdf},
DOI = {10.1111/j.1467-8659.2011.02041.x},

editor = {Bing-Yu Chen and Jan Kautz and Tong-Yee Lee and Ming C. Lin},

title = {Pacific Conference on Computer Graphics and Applications},

year = {2011},

isbn = {-},

issn = {1467-8659},

address = {Kaohsiung, Taiwan},

publisher = {Eurographics Association}
}

Slides

Presentation at Pacific Graphics, Sep 21^st 2011, Kaohsiung, Taiwan (pdf,pptx,video1,video2,video3)

Errata

Section 3.1.4 Entropy: missing closing bracket at the end of the formula.

Caption Figure 3: “inside the fusion algorithm”

Contact

Uwe Hahne (firstname.lastname @ tu-berlin.de)