Abstract
Traditionally most camera-based position estimation systems use only a few points to calibrate cameras. In this paper, we investigate a novel and alternate approach for 3D position estimation by using a larger number of points arranged in a 3D grid. We present an implementation of the active-space indexing mechanism which uses three cameras. Given the corresponding points in camera images, a precise estimation of the position can be obtained. The active-space indexing method can be also used as a spatial filter to eliminate the large number of possible corresponding pairs from consideration. This capability, unique only to the active-space indexing method, provides a tractable algorithm to the otherwise intractable situation.
Original language | English |
---|---|
Pages (from-to) | 135-150 |
Number of pages | 16 |
Journal | Graphical Models |
Volume | 63 |
Issue number | 3 |
DOIs | |
Publication status | Published - 2001 May |
Externally published | Yes |
Keywords
- 3D position estimation
- 3D spatial data structure
- Camera calibration for multiple camera images
- Computer vision techniques for 3D model building
- Geometric algorithms
- Virtual environments
ASJC Scopus subject areas
- Software
- Modelling and Simulation
- Geometry and Topology
- Computer Graphics and Computer-Aided Design