Adaptive imaging spectrometer in a time-domain filtering architecture

Abstract

Collecting a spectrum for each pixel in an image can give much useful information about a scene, such as chemical content, but generates a vast amount of data. Optical filters can select specific spectral features in advance, but are difficult to adapt in real time to different desired spectra. We have constructed an imaging spectrometer using a time-domain filtering architecture, capable of real-time spectral feature extraction and adaptation to different desired spectra. We demonstrate in real-time the abilities both to (i) recognize multiple specific colors in an image, and (ii) recognize individual colors while suppressing combinations of the same colors, an example of a sophisticated signal processing function that can be performed in this architecture.