Robust hand tracking by integrating appearance, location and depth cues

  • Authors:
  • Xiujuan Chai;Zhihao Xu;Qian Li;Bingpeng Ma;Xilin Chen

  • Affiliations:
  • Key Lab of Intelligent Information Processing of Chinese Academy of Sciences(CAS), Institute of Computing Technology, CAS Beijing, China;Huazhong University of Science and Technology Wuhan, China;Huazhong University of Science and Technology Wuhan, China;Huazhong University of Science and Technology Wuhan, China;Key Lab of Intelligent Information Processing of Chinese Academy of Sciences(CAS), Institute of Computing Technology, CAS Beijing, China

  • Venue:
  • Proceedings of the 4th International Conference on Internet Multimedia Computing and Service
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, a robust hand tracking algorithm is proposed for 3D sign language recognition application. The most challenging problems in hand tracking are the complex background, various illumination and articulated hand motion. In our work, based on the color and depth information simultaneously, two hands are well tracked by similarity optimization framework. However, in the fusion procedure, one important problem must be considered is that the data from the color and the depth channels are not always synchronous for the hardware reason. A two-layer difference comparison scheme is presented to determine whether the color and depth data are consistent. According to this consistency determination, the depth data can be used confidently or be deserted. Experiments on 300 sign language videos convincingly show the accuracy and robustness of the proposed hand tracking method. The visualized results also show the good performance even for the complex two-hand overlapping situations.