Smart mid-infrared metasurface microspectrometer gas sensing system.

Smart, low-cost and portable gas sensors are highly desired due to the importance of air quality monitoring for environmental and defense-related applications. Traditionally, electrochemical and nondispersive infrared (IR) gas sensors are designed to detect a single specific analyte. Although IR spectroscopy-based sensors provide superior performance, their deployment is limited due to their large size and high cost. In this study, a smart, low-cost, multigas sensing system is demonstrated consisting of a mid-infrared microspectrometer and a machine learning algorithm. The microspectrometer is a metasurface filter array integrated with a commercial IR camera that is consumable-free, compact ( ~ 1 cm3) and lightweight ( ~ 1 g). The machine learning algorithm is trained to analyze the data from the microspectrometer and predict the gases present. The system detects the greenhouse gases carbon dioxide and methane at concentrations ranging from 10 to 100% with 100% accuracy. It also detects hazardous gases at low concentrations with an accuracy of 98.4%. Ammonia can be detected at a concentration of 100 ppm. Additionally, methyl-ethyl-ketone can be detected at its permissible exposure limit (200 ppm); this concentration is considered low and nonhazardous. This study demonstrates the viability of using machine learning with IR spectroscopy to provide a smart and low-cost multigas sensing platform.

Human Interaction Understanding With Consistency-Aware Learning.

Compared with the progress made on human activity classification, much less success has been achieved on human interaction understanding (HIU). Apart from the latter task is much more challenging, the main causation is that recent approaches learn human interactive relations via shallow graphical representations, which are inadequate to model complicated human interactive-relations. This paper proposes a deep consistency-aware framework aiming at tackling the grouping and labelling inconsistencies in HIU. This framework consists of three components, including a backbone CNN to extract image features, a factor graph network to implicitly learn higher-order consistencies among labelling and grouping variables, and a consistency-aware reasoning module to explicitly enforcing consistencies. The last module is inspired by our key observation that the consistency-aware reasoning bias can be embedded into an energy function or a particular loss function, minimizing which delivers consistent predictions. An efficient mean-field inference algorithm is proposed, such that all modules of our network could be trained in an end-to-end fashion. Experimental results demonstrate that the two proposed consistency-learning modules complement each other, and both make considerable contributions in achieving leading performance on three benchmarks of HIU. The effectiveness of the proposed approach is further validated by experiments on detecting human-object interactions.

Genetic optimization of mid-infrared filters for a machine learning chemical classifier.

Miniaturized mid-infrared spectrometers present opportunities for applications that range from health monitoring to agriculture. One approach combines arrays of spectral filters with infrared photodetectors, called filter-array detector-array (FADA) microspectrometers. A paper recently reported a FADA microspectrometer in tandem with machine learning for chemical identification. In that work, a FADA microspectrometer with 20 filters was assembled and tested. The filters were band-pass, or band-stop designs that evenly spanned the microspectrometer's operating wavelength range. However, given that a machine learning classifier can be trained on an arbitrary filter basis, it is not apparent that evenly spaced filters are optimal. Here, through simulations with noise, we use a genetic algorithm to optimize six bandpass filters to best identify liquid and gaseous chemicals. We report that the classifiers trained with the optimized filter sets outperform those trained with evenly spaced filter sets and those handpicked to target the absorption bands of the chemicals investigated.

Mid-infrared spectral reconstruction with dielectric metasurfaces and dictionary learning.

Mid-infrared (MIR) spectroscopy has numerous industrial applications and is usually performed with Fourier-transform infrared (FTIR) spectrometers. While these work well for many purposes, there is currently much interest in alternative approaches that are smaller and lighter, i.e., MIR microspectrometers. Here we investigate all-dielectric metasurfaces as spectral filters for MIR microspectrometers. Two metasurface types are studied. For the first, we design, fabricate, and test a metasurface with a narrow and angularly tunable transmission stop band. We use it to reconstruct the transmission spectra of various materials. The second metasurface, investigated theoretically, possesses narrow passband features via symmetry-protected bound states in the continuum.

Vectorial Holograms with Spatially Continuous Polarization Distributions.

Metasurface-based holography presents opportunities for applications that include optical displays, data storage, and optical encryption. Holograms that control polarization are sometimes referred to as vectorial holograms. Most studies on this topic have concerned devices that display different images when illuminated with different polarization states. Fewer studies have demonstrated holographic images whose polarization varies spatially, i.e., as a function of the position within the image. Here, we experimentally demonstrate a vectorial hologram that produces an image with a spatially continuous distribution of polarization states, for the first time to our knowledge. An unlimited number of polarization states can be achieved within the image. Furthermore, the holographic image and its polarization map (polarization vs position in image) are independent. The same image can be thus encoded with different polarization maps. As far as we know, our approach is conceptually new. We anticipate that it could broaden the application scope of metasurface holography.

Mid- to long-wave infrared computational spectroscopy with a graphene metasurface modulator.

In recent years there has been much interest concerning the development of modulators in the mid- to long-wave infrared, based on emerging materials such as graphene. These have been frequently pursued for optical communications, though also for other specialized applications such as infrared scene projectors. Here we investigate a new application for graphene modulators in the mid- to long-wave infrared. We demonstrate, for the first time, computational spectroscopy in the mid- to long-wave infrared using a graphene-based metasurface modulator. Furthermore, our metasurface device operates at low gate voltage. To demonstrate computational spectroscopy, we provide our algorithm with the measured reflection spectra of the modulator at different gate voltages. We also provide it with the measured reflected light power as a function of the gate voltage. The algorithm then estimates the input spectrum. We show that the reconstructed spectrum is in good agreement with that measured directly by a Fourier transform infrared spectrometer, with a normalized mean-absolute-error (NMAE) of 0.021.

Detector-Only Spectrometer Based on Structurally Colored Silicon Nanowires and a Reconstruction Algorithm.

Spectroscopy is a cornerstone in the field of optics. Conventional spectrometers generally require two elements. The first provides wavelength selectivity, for example, diffraction grating or Michelson interferometer. The second is a detector (or detector array). Many applications would benefit from very small and lightweight spectrometers. This motivates us to investigate what may be regarded as an ultimate level of miniaturization for a spectrometer, in which it consists solely of a detector array. We demonstrate a chip containing 24 pixels, each comprising a silicon nanowire (Si NW) array photodetector formed above a planar photodetector. The NWs are structurally colored, enabling us to engineer the responsivity spectra of all photodetectors in the chip. Each pixel thus combines wavelength selectivity and photodetection functions. We demonstrate the use of our chip to reconstruct the spectrum of an unknown light source impinging upon it. This is achieved by an algorithm that takes as its inputs the measured photocurrents from the pixels and a library of their responsivity spectra.

Mid- to long-wave infrared computational spectroscopy using a subwavelength coaxial aperture array.

Miniaturized spectrometers are advantageous for many applications and can be achieved by what we term the filter-array detector-array (FADA) approach. In this method, each element of an optical filter array filters the light that is transmitted to the matching element of a photodetector array. By providing the outputs of the photodetector array and the filter transmission functions to a reconstruction algorithm, the spectrum of the light illuminating the FADA device can be estimated. Here, we experimentally demonstrate an array of 101 band-pass transmission filters that span the mid- to long-wave infrared (6.2 to 14.2 µm). Each filter comprises a sub-wavelength array of coaxial apertures in a gold film. As a proof-of-principle demonstration of the FADA approach, we use a Fourier transform infrared (FTIR) microscope to record the optical power transmitted through each filter. We provide this information, along with the transmission spectra of the filters, to a recursive least squares (RLS) algorithm that estimates the incident spectrum. We reconstruct the spectrum of the infrared light source of our FTIR and the transmission spectra of three polymer-type materials: polyethylene, cellophane and polyvinyl chloride. Reconstructed spectra are in very good agreement with those obtained via direct measurement by our FTIR system.

Experimental demonstration of infrared spectral reconstruction using plasmonic metasurfaces.

We computationally reconstruct short- to long-wave infrared spectra using an array of plasmonic metasurface filters. We illuminate the filter array with an unknown spectrum and measure the optical power transmitted through each filter with an infrared microscope to emulate a filter-detector array system. We then use the recursive least squares method to determine the unknown spectrum. We demonstrate our method with light from a blackbody. We also demonstrate it with spectra generated by passing the light from the blackbody through various materials. Our approach is a step towards miniaturized spectrometers spanning the short- to long-wave infrared based on filter-detector arrays.