Lazy Camera Calibration Toolbox Development

We offer a guided research project in the field of computer vision and robotics, which aims to develop a new efficient camera calibration toolbox for monocular/stereo/RGB-D sensors.

Motivation

Camera calibration is a basic topic in the vision community where many methods/tools were created to calibrate our cameras. In those tools, a single checkboard is usually used, which has a robust and precise performance if the checkboard is printed on an absolute planar region. Otherwise, the accuracy of calibration will be reduced.

Imagine if you don't want to paste a dirty checkerboard paper (like the third figure) on a perfectly planar surface, maybe a lazy camera calibration toolbox is good for you. Via the idea of Manhattan World (MW) and vanishing points, the toolbox still can give you accurate intrinsic parameters and does not need you to do too much.

Resources

• Matlab TOOLBOX_calib: http://www.vision.caltech.edu/bouguetj/calib_doc/
• Zhang, Zhengyou. "A flexible new technique for camera calibration." TPAMI 2000.
• Geiger, Andreas, et al. "Automatic camera and range sensor calibration using a single shot." ICRA 2012

Requirements

• Familiar with Multi View Geometry.
• Good C++/Matlab skill.

If you are interested in any of these topics, please contact us via e-mail.

Yanyan Li

Federico Tombari