Share Email Print

Optical Engineering

Online optimal path decoder of hidden Markov model and its application to connected gesture recognition
Author(s): Monalisa Mazumdar; Mun-Ho Jeong; Bum-Jae You
Format Member Price Non-Member Price
PDF $20.00 $25.00
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

We model a recognition problem for connected hand gestures to find an optimal path through a hidden Markov model (HMM) directed acyclic graph. To determine this optimal path, an online graph search method is proposed that decodes the observed gesture pattern and evaluates the optimal graph node at each time frame of the continuously deepening HMM graph. The temporal characteristic of gesture recognition is subsequently handled by introducing a rejection threshold time that acts as a depth-wise sliding window for pruning unnecessary graph nodes. The functional depth of the graph is defined by this depth rejection threshold. Experimental comparison of our algorithm with other HMM-based search algorithms demonstrates the effectiveness and robustness of our method.

Paper Details

Date Published: 1 August 2008
PDF: 12 pages
Opt. Eng. 47(8) 087204 doi: 10.1117/1.2969123
Published in: Optical Engineering Volume 47, Issue 8
Show Author Affiliations
Monalisa Mazumdar, Korea Institute of Science and Technology (Korea, Republic of)
Mun-Ho Jeong, Korea Institute of Science and Technology (Korea, Republic of)
Bum-Jae You, Korea Institute of Science and Technology (Korea, Republic of)

© SPIE. Terms of Use
Back to Top