View Invariant Gesture Recognition using 3D Motion Primitives

Michael Boelstoft Holte; Thomas B. Moeslund

doi:10.1109/ICASSP.2008.4517730

View Invariant Gesture Recognition using 3D Motion Primitives

Michael Boelstoft Holte, Thomas B. Moeslund

Research output: Contribution to journal › Conference article in Journal › Research › peer-review

21 Citations (Scopus)

Abstract

This paper presents a method for automatic recognition of human gestures. The method works with 3D image data from a range camera to achieve invariance to viewpoint. The recognition is based solely on motion from characteristic instances of the gestures. These instances are denoted 3D motion primitives. The method extracts 3D motion from range images and represent the motion from each input frame in a view invariant manner using harmonic shape context. The harmonic shape context is classified as a 3D motion primitive. A sequence of input frames results in a set of primitives that are classified as a gesture using a probabilistic edit distance method. The system has been trained on frontal images (0deg camera rotation) and tested on 240 video sequences from 0deg and 45deg. An overall recognition rate of 82.9% is achieved. The recognition rate is independent of the viewpoint which shows that the method is indeed view invariant.

Original language	English
Journal	Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing
Pages (from-to)	797-800
ISSN	1520-6149
DOIs	https://doi.org/10.1109/ICASSP.2008.4517730
Publication status	Published - 2008
Event	IEEE International Conference on Acoustics, Speech, and Signal Processing - Las Vegas, United States Duration: 31 Mar 2008 → 4 Apr 2008

Conference

Conference	IEEE International Conference on Acoustics, Speech, and Signal Processing
Country/Territory	United States
City	Las Vegas
Period	31/03/2008 → 04/04/2008

Access to Document

10.1109/ICASSP.2008.4517730

AUB Link

Search for the material in Aalborg University Library's search engine

Cite this

@inproceedings{738ed5c06ad211dd92a2000ea68e967b,

title = "View Invariant Gesture Recognition using 3D Motion Primitives",

abstract = "This paper presents a method for automatic recognition of human gestures. The method works with 3D image data from a range camera to achieve invariance to viewpoint. The recognition is based solely on motion from characteristic instances of the gestures. These instances are denoted 3D motion primitives. The method extracts 3D motion from range images and represent the motion from each input frame in a view invariant manner using harmonic shape context. The harmonic shape context is classified as a 3D motion primitive. A sequence of input frames results in a set of primitives that are classified as a gesture using a probabilistic edit distance method. The system has been trained on frontal images (0deg camera rotation) and tested on 240 video sequences from 0deg and 45deg. An overall recognition rate of 82.9% is achieved. The recognition rate is independent of the viewpoint which shows that the method is indeed view invariant.",

author = "Holte, {Michael Boelstoft} and Moeslund, {Thomas B.}",

year = "2008",

doi = "10.1109/ICASSP.2008.4517730",

language = "English",

pages = "797--800",

journal = "Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing",

issn = "1520-6149",

publisher = "IEEE Signal Processing Society",

note = "IEEE International Conference on Acoustics, Speech, and Signal Processing ; Conference date: 31-03-2008 Through 04-04-2008",

}

TY - GEN

T1 - View Invariant Gesture Recognition using 3D Motion Primitives

AU - Holte, Michael Boelstoft

AU - Moeslund, Thomas B.

PY - 2008

Y1 - 2008

N2 - This paper presents a method for automatic recognition of human gestures. The method works with 3D image data from a range camera to achieve invariance to viewpoint. The recognition is based solely on motion from characteristic instances of the gestures. These instances are denoted 3D motion primitives. The method extracts 3D motion from range images and represent the motion from each input frame in a view invariant manner using harmonic shape context. The harmonic shape context is classified as a 3D motion primitive. A sequence of input frames results in a set of primitives that are classified as a gesture using a probabilistic edit distance method. The system has been trained on frontal images (0deg camera rotation) and tested on 240 video sequences from 0deg and 45deg. An overall recognition rate of 82.9% is achieved. The recognition rate is independent of the viewpoint which shows that the method is indeed view invariant.

AB - This paper presents a method for automatic recognition of human gestures. The method works with 3D image data from a range camera to achieve invariance to viewpoint. The recognition is based solely on motion from characteristic instances of the gestures. These instances are denoted 3D motion primitives. The method extracts 3D motion from range images and represent the motion from each input frame in a view invariant manner using harmonic shape context. The harmonic shape context is classified as a 3D motion primitive. A sequence of input frames results in a set of primitives that are classified as a gesture using a probabilistic edit distance method. The system has been trained on frontal images (0deg camera rotation) and tested on 240 video sequences from 0deg and 45deg. An overall recognition rate of 82.9% is achieved. The recognition rate is independent of the viewpoint which shows that the method is indeed view invariant.

U2 - 10.1109/ICASSP.2008.4517730

DO - 10.1109/ICASSP.2008.4517730

M3 - Conference article in Journal

SN - 1520-6149

SP - 797

EP - 800

JO - Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing

JF - Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing

T2 - IEEE International Conference on Acoustics, Speech, and Signal Processing

Y2 - 31 March 2008 through 4 April 2008

ER -