Implementation of easy to use and accurate calibration method for RGB-D camera - ON-218

Preferred Disciplines: Computer Science or Engineering, Computer Graphics (Bachelor degree or above)
Company: DoubleMe
Project Length: 4-6 months
Desired start date: As soon as possible
Location: Toronto, ON
No. of Positions: 1-2
Preferences: None

About the Company: 

We are a startup with real-time non-rigid 3D reconstruction solutions and services based on single RGB-D.

Project Description:

We hope to get simple, fast and more accurate calibration software implementation with some facilities for general commercial RGB-D cameras such as Kinect/Kinect2, RealSense, Xtion Pro Live, Astro Pro.

Research Objectives:

  • Need to support alignment/registration on the basis of Depth frame data (In OpenNI, their API uses Depth to RGB regitration on grabbing each frames)
  • Need to support to get geometric calibration data, such as intrinsics for RGB and Depth/IR sensors, extrinsics between RGB and Depth/IR sensors (refer to OpenCV's calibration API)
  • Need to support easy steps from start of calibration capture to final calibration data (e.g: Start calibration software --> perform simple action with checkerboard or something in front of RGB-D camera --> process/calculate --> get final calibration data(in forms of transformation matrix))
  • Need to support different resolution of RGB and Depth/IR sensors
  • For ease of use, prefer to support GUI for controls
  • Whole calibration steps should be completed in 10~20 minutes
  • Use(or without use) of some materials(like a chessboard) in calibration capture stage
  • Use general know algoritms in calibration process of single sensor or stereo calibration (we prefer to use OpenCV)
  • RMS errors should not be exceeded 0.3~0.4 in calibration stage
  • Reprojection errors should not be exceeded 0.5~0.6 in final calibration stage

Methodology:

  • Recommend to adopt some Agile approach and we hope to see result of research/development on each sprint/milestone
  • Need to get overall planning on start of this project
  • Recommend to use our SCM/IssueTrack system in whole development phases

Expertise and Skills Needed:

  • Bachelor degree or above preferred in Computer Vision or  Computer Graphics
  • Should have some experiences in Computer Vision fundamentals for camera calibration algorithms
  • Should have experiences in programming with C/C++ languages under Windows, Linux OS.

For more info or to apply to this applied research position, please

  1. Check your eligibility and find more information about open projects
  2. Interested students need to get the approval from their supervisor and send their CV along with a link to their supervisor’s university webpage by applying through the webform.
Program: