< Back to previous page

Project

Tiny Thin-film Filters from a Different Angle: Correcting Angular Dependency for Spectral Imaging

Spectral imaging is a technology that combines photography and
spectroscopy by sampling the electromagnetic spectrum for each point in the scene. This spectrum can be used as a ``fingerprint'' to identify different materials and analyze their properties.


Recent developments have enabled the integration of thin-film filters onto pixels of an image sensor, making it possible for pixels to sample specific wavelengths.Furthermore, the filter-on-chip integration makes spectral cameras robust, compact, and lightweight enough to be used in various applications including remote sensing, industrial quality control, healthcare and precision agriculture.
However, to enable mass production and widespread adoption across industries, it will be essential to minimize costly and time-consuming expert support for camera calibration.


One of the key concerns in calibration is the angle-dependent transmittance of thin-film filters.
When illuminated by focused light, the filters might select other wavelengths than originally intended, causing the measured spectra to be smoothed and spectrally shifted. This can strongly impact application performance and hence also limits the selection of compatible lenses.
Hardware solutions include the use of filter materials that are less angle sensitive, expensive telecentric lenses, or the development of lens-specific filter designs.
These solutions cannot always be implemented due to a lack of compatible materials or budgetary constraints.

Instead, we propose a software correction approach that enables cost-efficient use of the same sensor with most commercially available non-telecentric lenses.
In this regard, this work has two major contributions.


The first contribution consists of new and insightful analytical models that can be used to predict the transmittance of thin-film filters illuminated by camera lenses. The models cover arbitrarily tilted circular apertures and lenses with vignetting.
Using analytical approximations predicated on classical thin-film theory, spectrally shifted measurements can be corrected in a practical way. As a result, it does not matter anymore which lens is used as long as the shift can be predicted and other distortions are negligible.

The second contribution is the derivation of a novel thin-film filter model which takes into account the finite width of integrated thin-film filters. This ``tiny filter'' model predicts significant deviations from classical thin-film theory which assumes an infinite filter width.
The predicted deviations are observed in multiple spectral cameras and result in the formulation of new limitations for filter miniaturization. Furthermore, the model might be used to build software tools to efficiently design tiny filters or perhaps avoid unnecessary investments.

In conclusion, the developed software corrections expand the selection of compatible lenses and simplify the calibration process as long as classical thin-film theory applies.
When this is no longer the case, no correction methods currently exist. The development of such methods is therefore suggested as a topic for further investigations.
 

Date:17 Oct 2016 →  13 Oct 2020
Keywords:hyperspectral, multispectral, thin-film, thin film
Disciplines:Nanotechnology, Design theories and methods, Optics, electromagnetic theory
Project type:PhD project