Implementation and Analysis of Algorithms for Spectral Reconstruction Based on a Multispectral Sensor with Integrated Plasmonic Filters

Abstract: In the near future, microspectrometers will play an important role in various field of application. By integrating plasmonic filter technology it is possible to reduce costs and time of production as well as to improve the versatile usability of microspectrometers.

Specific filter properties cause a sensor response to suboptimally represent the incident spectrum. The main issue is to reconstruct the incident spectrum with suitable algorithms.

In the scope of this thesis multiple approaches for solving the linear, inverse problem are examined and implemented using the programming language Phython. Besides simple algorithms of linear algebra and a regularization, the approach was to map the real sensor properties to optimum ones. Simulations with various filter curves were performed to analyse and rate the use algorithms.

Finally, the spectra of real LEDs are reconstructed from measured values of the sensor validating the realized algorithms.