Objectives

Light field cameras, like any other camera, require calibration before they can be used for 3D reconstruction purposes. These cameras have zoom and focus settings which help optimize for the scene under study. However, these settings mean the camera has to be calibrated for every set of zoom and focus settings chosen. This can be tedious since each calibration requires capturing 15 images, and takes hours with our current set-up. Furthermore, the camera’s interface makes it difficult to match zoom and focus settings to a previous calibration, since it does not show these values.

Investigate a calibration model that isn’t specific to the camera’s zoom and focus settings. Ideally, the calibration model generates a normal light field camera model [Dansereau13] for each zoom/focus settings, so that it can be readily used with current light field reconstruction techniques. It is also desirable that the calibration require a lower amount of images than it would take to calibrate the camera for each zoom/focus settings covered.

The work-plan will require acquiring knowledge on camera modeling and calibration, specific to light field cameras, and possibly on optimization techniques used for these purposes. It should also include a study of the state of the art in this area such as [Atienza01].

References:

[Dansereau13] Dansereau, Donald G., Oscar Pizarro, and Stefan B. Williams. "Decoding, calibration and rectification for lenselet-based plenoptic cameras." Computer Vision and Pattern Recognition (CVPR), 2013 IEEE Conference on. IEEE, 2013.

[Atienza01] Atienza, Rowel, and Alex Zelinsky. "A practical zoom camera calibration technique: an application on active vision for human-robot interaction." Proc. Australian Conference on Robotics and Automation. 2001.

More information:

http://users.isr.ist.utl.pt/~jag/msc/msc_2018_2019.html

Requirements (grades, required courses, etc):

Interest in imaging sensors.

Place for conducting the work-proposal:

ISR / IST