An improved colored-marker based registration method for AR applications

  • Authors:

  • Xiaowei Li;Yue Liu;Yongtian Wang;Dayuan Yan;Dongdong Weng;Tao Yang

  • Affiliations:

  • School of Information Science and Technology, Beijing Institute of Technology, Beijing;School of Information Science and Technology, Beijing Institute of Technology, Beijing;School of Information Science and Technology, Beijing Institute of Technology, Beijing;School of Information Science and Technology, Beijing Institute of Technology, Beijing;School of Information Science and Technology, Beijing Institute of Technology, Beijing;School of Information Science and Technology, Beijing Institute of Technology, Beijing

  • Venue:
  • ICCSA'05 Proceedings of the 2005 international conference on Computational Science and Its Applications - Volume Part III
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

Registration is crucial in an Augmented Reality (AR) system for it determines the performance of alignment between virtual objects and real scene. Colored-makers with known world coordinates are usually put in the target scene beforehand to help get a real-time, precise registration because they can provide explicit 3D/2D correspondences and four such correspondences can produce adequate and accurate equations of the pose matrix if the camera's intrinsic matrix has already been calibrated, and then registration can be achieved by solving these equations. However, usually only limited number of (e.g. two or three) markers out of four can be captured and this will make the colored-marker based method fail. In order to overcome such shortcomings an improved colored-marker based registration method is proposed in this paper which works when the target scene is a plane. The proposed method integrates both 3D/2D and 2D/2D information by updating the cost function used in the optimization step of RANSAC, and thus combines the virtues of homography based method. Experimental result shows that the proposed method can provide acceptable pose estimation and its potential to be applied in actual AR systems.