A Pseudo-Sample Based Low Cost 3D Human Motion Capture System Using Wireless Camera Sensor Network

다국어 초록 (Multilingual Abstract)

There is a growing interest in 3D contents in the global market
following the recent developments in 3D movies, 3D TVs, and 3D
smartphones.
However, the 3D content creation is still dominated by professionals
due to the high cost of 3D motion capture instruments.
The availability of a low-cost motion capture system will promote 3D
content generation by general users and accelerate the growth of the
3D market.
In this paper, we propose a real-time low-cost motion capture system,
called Samba.
The Samba motion capture system is based on a portable low-cost
wireless camera sensor network and it can reconstruct accurate 3D
full-body poses at 16 frames per second using only eight markers on
the subject body.
The processing times on camera motes are significantly reduced by
using a smaller number of markers.
A pseudo-pose based data-driven 3D human pose reconstruction method is
proposed to reduce the required computation time and to improve the 3D
reconstruction accuracy.
The performance of Samba is evaluated extensively in experiments.

목차 (Table of Contents)

• 1 INTRODUCTION
• 2 Related works
• 3 An Overview of the Samba Motion Capture System
• 4 Data-Driven Human Pose Estimation
• 4.1 Local Pose Space Method
• 4.2 Pseudo-Pose Generation
• 4.2.1 One-DOF Joints
• 4.2.2 Two-DOF Joints
• 4.2.3 Initial Values for Optimization
• 5 System Implementation
• 5.1 Camera Mote
• 5.2 Samba Server
• 5.3 Human Pose Database
• 5.4 Marker Detection and Tracking
• 5.5 3D Reconstruction of Markers
• 6 Experimental Results
• 6.1 Image Processing Times at a Camera Mote
• 6.2 Number of Neighbors(K)
• 6.3 Number of Pseudo-Poses(N)
• 6.4 CMU Motion Capture Database
• 6.5 Demonstration of the Samba Motion Capture System
• 7 Conclusion Abstract (In Korean)