Sensing and compressing 3-D models
- Sandia National Labs., Albuquerque, NM (United States). Intelligent System Sensors and Controls Dept.
The goal of this research project was to create a passive and robust computer vision system for producing 3-D computer models of arbitrary scenes. Although the authors were unsuccessful in achieving the overall goal, several components of this research have shown significant potential. Of particular interest is the application of parametric eigenspace methods for planar pose measurement of partially occluded objects in gray-level images. The techniques presented provide a simple, accurate, and robust solution to the planar pose measurement problem. In addition, the representational efficiency of eigenspace methods used with gray-level features were successfully extended to binary features, which are less sensitive to illumination changes. The results of this research are presented in two papers that were written during the course of this project. The papers are included in sections 2 and 3. The first section of this report summarizes the 3-D modeling efforts.
- Research Organization:
- Sandia National Labs., Albuquerque, NM (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- AC04-94AL85000
- OSTI ID:
- 573305
- Report Number(s):
- SAND--98-0503; CONF-960672--3; CONF-970679--2; ON: DE98002851
- Country of Publication:
- United States
- Language:
- English
Similar Records
Automated dimensional analysis using a light-sectioning microscope
Automation recognition of tracking of objects. Final report, 1 January 1985-31 December 1986