Share Email Print
cover

Proceedings Paper

Human motion recognition based on features and models selected HMM
Author(s): Haixiang Lu; Hongjun Zhou
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

This paper research on the motion recognition based on HMM with Kinect. Kinect provides skeletal data consist of 3D body joints with its lower price and convenience. In this work, several methods are used to determine the optimal subset of features among Cartesian coordinates, distance to hip center, velocity, angle and angular velocity, in order to improve the recognition rate. K-means is used for vector quantization and HMM is used as recognition method. HMM is an effective signal processing method which contains time calibration, provides a learning mechanism and recognition ability. Cluster numbers of K-means, structure and state numbers of HMM are optimized as well. The proposed methods are applied to the MSR Action3D dataset. Results show that the proposed methods obtain better recognition accuracy than the state of the art methods.

Paper Details

Date Published: 4 March 2015
PDF: 6 pages
Proc. SPIE 9521, Selected Papers from Conferences of the Photoelectronic Technology Committee of the Chinese Society of Astronautics 2014, Part I, 952107 (4 March 2015); doi: 10.1117/12.2087230
Show Author Affiliations
Haixiang Lu, Tongji Univ. (China)
Hongjun Zhou, Tongji Univ. (China)


Published in SPIE Proceedings Vol. 9521:
Selected Papers from Conferences of the Photoelectronic Technology Committee of the Chinese Society of Astronautics 2014, Part I
Xun Hou; Zhihong Wang; Lingan Wu; Jing Ma, Editor(s)

© SPIE. Terms of Use
Back to Top