3D hand posture recognition using multicam

Mohd Nazrin , Muhammad and Nurdiana, Nordin (2011) 3D hand posture recognition using multicam. In: 2011 IEEE International Conference on Signal and Image Processing Applications {ICSIPA 2011), 16-18 November 2011, Hotel Maya, Kuala Lumpur. (Submitted)

[img] Text
3D hand posture recognition using multicam251.pdf

Download (937kB)


This paper presents the hand posture recognition in 3D using the MultiCam, a monocular 2D/3D camera developed by Center of Sensorsystems (ZESS). The :VlultiCam is a camera which is capable to provide high resolution of color data acquired from CMOS sensors and low resolution of distance (or range) data calculated based on timeof- flight (ToF) technology using Photonic Mixer Device (PMD) sensors. The availability of the distance data allows the hand posture to be recognized in z-axis direction without complex computational algorithms which also enables the program to work in real-time processing as well as eliminates the background effectively. The hand posture recognition will employ a simple but robust algorithm by checking the number of fingers detected around virtually created circle centered at the Center of Mass (CoM) of the hand and therefore classifies the class associated with a particular hand posture. At the end of this paper, the technique that uses intersection between the circle and fingers as the method to classify the hand posture which entails the MultiCam capability is proposed. This technique is able to solve the problem of orientation, size and distance invariants by utilizing the distance data.

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: multicam, hand posture, hand posture recognition, real-time processing, 3D, distance data
Subjects: T Technology > TR Photography
Divisions: Faculty of Manufacturing Engineering
Depositing User: Noor Rahman Jamiah Jalil
Date Deposited: 21 Oct 2015 09:13
Last Modified: 21 Oct 2015 09:14
URI: http://eprints.utem.edu.my/id/eprint/15084
Statistic Details: View Download Statistic

Actions (login required)

View Item View Item