| The demand for miniaturized optical spectrometer systems is steadily increasing, for applications such as biochemical material identification, atmosphere monitoring, and military target tracking. Unfortunately, sensing front-ends typically generate a burdensome amount of data that must be processed by large computing systems; the required processing back-ends often limit the compactness and efficiency of complete systems. Ideally, we would design new devices that enable easy implementation of data filtering in a miniaturized front-end, with the capability to adapt to a wide range of sensing tasks.; Here we present a new micro spectrometer and compatible information processing system, for applications requiring a flexible, portable, complete sensing system. The standing-wave spectrometer enjoys the advantages of a traditional Fourier transform spectrometer, including spectral multiplexing and simple interferogram output; however it has a compact, linear optical design requiring only two components, a partially transmitting photodetector and a movable mirror. We discuss development of a low power, continuously scanning, miniature Si mirror-actuator, and Si and GaAs photodiodes with thin active regions, for visible and near-infrared devices as small as 17 x 13 x 1 mm. We demonstrate spectral resolution of 100 cm-1 (4 run at lambda = 633 nm), and the unique ability to adapt resolution in real time to optimize signal-to-noise ratio.; For spectral data processing, we present a new time-domain filtering concept that minimizes computing requirements. For discrimination among a set of known spectra, we directly use the interferograms generated by a Fourier transform spectrometer to calculate inner products in the time domain. Our method efficiently uses prior knowledge of the known spectra to relax data handling requirements, typically by factors of 10--100, enabling real-time spectra-selective imaging. We can also directly identify optical source types by their spectral bandwidth, or temporal coherence, without prior knowledge of other wavelength-based characteristics. With simple processing we demonstrate coherence-selective imaging, enabling a system to discriminate among broadband sources, LEDs, lasers, or other spectrally narrow reflections in a 2D scene.; Integration of these optical microspectrometers and spectral data processing methods could enable a new generation of portable sensing systems. |
PDF Full Text Request