PPT Slide

3D Mapping Methodology

Stanford University School of Engineering

Department of Mechanical Engineering

XX February 2000

Left camera

Right camera

Image Capture System

3D Processing

Radially distorted image

Radially distorted image

L/R Lens properties

Imaging geometry

Distortion correction algoritm

Distortion-free image

Distortion-free image

Distortion-free image/L

Distortion-free image/R

3D Mesh

• Known mesh vs. camera position
• Quantifiable object measurements with known error

(Pinhole Camera Model)

Previous slide

Next slide

Back to first slide

View graphic version

Mention:

• 1 vs. 2 cameras (current limits in internal position knowledge)
  • Doppler based navigation (1cm)

• 2 cameras gets you down to machining errors mils (will show why this important in geometrical derivation next)

• Individual Camera Calibration
  • Model development from calibration process
    • Calculated intrinsic and extrinsic parameters fed into stereo? pipeline for mesh building.
  • Distortion free images (de-warped) (done)

• Stereo Pipeline
  • Develop calibration process to characterize errors based on mesh-building algorithm.
  • Laser assist?
  • Calibration process would also allow for using other “pipelines”

• Mapping/mesh-stitching
  • Calibrate system (ROV) wide errors using navigation data and correcting using mesh stitching algorithms.