Open Collections

Acquisition of transparent refractive media

University of British Columbia

Transparent refractive media are invisible but for the distortions they impart upon a background scene. Computerised acquisition of such media can therefore often not be performed via traditional scanning methods. By capturing refracted backgrounds rather than reflections off the target media itself, we develop techniques for reconstructing the intervening refractive index distribution for both static and time-varying media. The approach is based on tracking optical distortions and then performing tomographic reconstruction. For multi-view tomography we first require a suitably calibrated camera array. To this end we show how to temporally synchronise and geometrically calibrate an array of consumer-grade video cameras that can scale to larger sizes, and at lower cost, than a comparative array of machine vision cameras. For media of low dynamic refractive index range, such as mixing gases, we show how to acquire data and formulate a linear least-squares problem to solve for the refractive index distribution. Unlike traditional methods of fluid flow measurement, ours is non-invasive and fully volumetric. For materials of higher dynamic refractive index range, we develop an alternative acquisition method based on temporally-encoded structured light patterns. Media causing significant distortion of light rays give rise to a large, nonlinear inverse problem. Results indicate that grid resolution relative to the minimum refractive feature size is a key factor limiting the accuracy of reconstructions.

Text

2012-11-28

Attribution-NonCommercial 3.0 Unported

http://hdl.handle.net/2429/43621

Computer Science

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc/3.0/