Spatiotemporal Phase Congruency Based Invariant Features for Human Behavior Classification

Authors:
Hao Wang;Yi Xu;Xiaokang Yang;Li Song;Wenjun Zhang
Affiliations:
Institute of Image Communication and Information Processing, and Shanghai Key Laboratory of Digital Media Processing and Transmissions,;Institute of Image Communication and Information Processing, and Shanghai Key Laboratory of Digital Media Processing and Transmissions,;Institute of Image Communication and Information Processing, and Shanghai Key Laboratory of Digital Media Processing and Transmissions,;Institute of Image Communication and Information Processing, and Shanghai Key Laboratory of Digital Media Processing and Transmissions,;Institute of Image Communication and Information Processing, and Shanghai Key Laboratory of Digital Media Processing and Transmissions,
Venue:
PCM '09 Proceedings of the 10th Pacific Rim Conference on Multimedia: Advances in Multimedia Information Processing
Year:
2009

Citing 6
Cited 0

Feature detection from local energy

Pattern Recognition Letters
Computer and Robot Vision

Computer and Robot Vision
Latent dirichlet allocation

The Journal of Machine Learning Research
Recognizing Human Actions: A Local SVM Approach

ICPR '04 Proceedings of the Pattern Recognition, 17th International Conference on (ICPR'04) Volume 3 - Volume 03
On Space-Time Interest Points

International Journal of Computer Vision
Behavior recognition via sparse spatio-temporal features

ICCCN '05 Proceedings of the 14th International Conference on Computer Communications and Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

In applications of behavior recognition, the use of spatiotemporal invariant feature points can improve the robustness to noise, illumination and geometric distortions. In this paper, we develop a novel detection model of spatiotemporal invariant feature by generalizing the notion of image phase congruency to video volume phase congruency. The proposed model detects feature points by measuring the spatiotemporal phase congruency of Fourier series components along with their characteristic scale and principal orientation. Compared with other state-of-the-art methods, the key advantages of this interest point detector include the invariance to contrast variations and more precise feature location. Furthermore, an invariant feature descriptor is advanced based on the phase congruency map, resulting in enhanced discriminative power in classification tasks. Experimental results on KTH human motion dataset demonstrate the validity and effectiveness of the extracted invariant features in the human behavior recognition scheme.