Online smoothing for markerless motion capture

Bodo Rosenhahn, Thomas Brox, Daniel Cremers, Hans Peter Seidel

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Scopus citations


Tracking 3D objects from 2D image data often leads to jittery tracking results. In general, unsmooth motion is a sign of tracking errors, which, in the worst case, can cause the tracker to loose the tracked object. A straightforward remedy is to demand temporal consistency and to smooth the result. This is often done in form of a post-processing. In this paper, we present an approach for online smoothing in the scope of 3D human motion tracking. To this end, we extend an energy functional by a term that penalizes deviations from smoothness. It is shown experimentally that such online smoothing on pose parameters and joint angles leads to improved results and can even succeed in cases, where tracking without temporal consistency assumptions fails completely.

Original languageEnglish
Title of host publicationPattern Recognition - 29th DAGM Symposium, Proceedings
PublisherSpringer Verlag
Number of pages10
ISBN (Print)3540749330, 9783540749332
StatePublished - 2007
Externally publishedYes
Event29th Annual Symposium of the Deutsche Arbeitsgemeinschaft fur Mustererkennung, DAGM 2007 - Heidelberg, Germany
Duration: 12 Sep 200714 Sep 2007

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume4713 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference29th Annual Symposium of the Deutsche Arbeitsgemeinschaft fur Mustererkennung, DAGM 2007


Dive into the research topics of 'Online smoothing for markerless motion capture'. Together they form a unique fingerprint.

Cite this