electronic visualization laboratory, university of illinois a Camera Based Automatic Calibration for the Varrier™ System Jinghua Ge, Dan Sandin, Tom Peterka, Todd Margolis, Tom DeFanti Electronic Visualization Laboratory, University of Illinois at Chicago
6
Embed
Electronic visualization laboratory, university of illinois at chicago Camera Based Automatic Calibration for the Varrier™ System Jinghua Ge, Dan Sandin,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
electronic visualization laboratory, university of illinois at chicago
Camera Based Automatic Calibration for the Varrier™ System
Jinghua Ge, Dan Sandin, Tom Peterka, Todd Margolis, Tom DeFanti
Electronic Visualization Laboratory, University of Illinois at Chicago
electronic visualization laboratory, university of illinois at chicago
Introduction of Varrier
Varrier is a head-tracked, 35 panel tiled autostereoscopic VR display
Combination of physical and virtual linescreens to achieve autostereo
Virtual linescreen registration parameters:
Pitch, z rotation, optical thickness, x shift
electronic visualization laboratory, university of illinois at chicago
Camera automated calibration
Two video cameras simulate human eyes
Goal:
left eye sees left image,
right eye sees right image
Calibration Patterns:
Color pattern: left eye red, right eye blue
Cross bar pattern: white cross bars, left eye 45º , right eye -45º
electronic visualization laboratory, university of illinois at chicago
Calibration Algorithms
• Heuristic rough tune– Color pattern enhances
moire bars– Register rotaton until moire
bar parallel to linescreen– Register optical thickness
until no bar in at least one eye image
– Register shift until left eye see red, right eye see blue
electronic visualization laboratory, university of illinois at chicago
Calibration Algorithms (cond’)
• Adaptive fine tune– Cross bar pattern enhances brightness/ghost– 2D searching for maximum brightness-ghost– Multiple fine tune get better precision
electronic visualization laboratory, university of illinois at chicago
Performance and Conclusion
• Heuristic rough tune achieve close result very fast– Usually in less than 70 steps
• Multiple adaptive fine tune get best parameter set in high precision– Usually less than 100 steps
• Registration runs at 3fps. One minute for each screen.
• One on-axis registration works for free on-axis movement