3d_camera

This repository is a set of tools and workflows for doing stereoscopic 3D reconstruction on a Raspberry Pi (3).

There are also tools for general stereoscopy and our process will be documented so that others may learn and make progress from our findings.

This project was done as a Senior Design Project for Electrical & Computer Engineering Undergraduate Program at the University of Pittsburgh by Aaron Paczak, Ben Friedman, and Sam Nosenzo.

Our Hardware setup:

  • Raspberry Pi 3
  • 2 x Pi NoIR v2 Cameras
  • Arducam multi camera Adapter Optional
  • outside button circuit
  • Canon Rebel t1i (or any camera compatible with the gphoto library)
    • We also did registration of the original image and the DSLR photo

Dependencies (all available on pip)

For OpenCV:

  • python-opencv
  • numpy
  • matplotlib
  • stereovision (by erget)

For Raspberry Pi/Hardware setup

  • raspberry-gpio-python (we needed this for compatibility with the Arducam multi-camera adapter)
  • What we should've used was probably the picamera pip package

Improvements for this repository:

  • We Should've had better calibration images for the Pi
  • Should've used a better checkerboard print for calibration
  • Can't control camera parameters as we would like to using the pi cameras
  • Can't take images simulataneously on the Pi due to arducam setup

TODO

  • put in pictures of our hardware setup
  • Explain file directory and structure