Comparative discrimination spectral detection system and method for the identification of chemicals with overlapping spectral signatures
Inventors
Poutous, Menelaos K. • Aggarwal, Ishwar D. • Major, Kevin J. • Sanghera, Jas S. • Ewing, Ken
Assignees
University of North Carolina at Charlotte • US Department of Navy
Publication Number
US-9857295-B2
Publication Date
2018-01-02
Expiration Date
2035-12-11
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
A comparative discrimination spectral detection (CDSD) system for the identification of chemicals with overlapping spectral signatures, including: a radiation source for delivering radiation to a sample; a radiation collector for collecting radiation from the sample; a plurality of beam splitters for splitting the radiation collected from the sample into a plurality of radiation beams; a plurality of low-resolution optical filters for filtering the plurality of radiation beams; a plurality of radiation detectors for detecting the plurality filtered radiation beams; and a processor for: receiving a set of reference spectra related to a set of target chemicals and generating a set of base vectors for the set of target chemicals from the set of reference spectra, wherein the set of base vectors define a geometrical shape in a configuration space; receiving a set of filtered test spectra from the plurality of radiation detectors and generating a set of test vectors in the configuration space from the set of filtered test spectra; assessing a geometrical relationship of the set of test vectors and the geometrical shape defined by the set of base vectors in the configuration space; and based on the assessed geometrical relationship, establishing a probability that a given test spectrum or spectra matches a given reference spectrum or spectra.
Core Innovation
The invention is a comparative discrimination spectral detection (CDSD) system and method for the identification of chemicals with overlapping spectral signatures. The CDSD system includes a radiation source to deliver radiation to a sample, a radiation collector to collect radiation from the sample, beam splitters to split the collected radiation into multiple beams, low-resolution optical filters to filter the radiation beams, radiation detectors to detect the filtered beams, and a processor. This processor receives reference spectra for target chemicals, generates base vectors defining a geometrical shape in a configuration space, receives filtered test spectra, generates test vectors, assesses the geometrical relationship between the test vectors and base vectors, and establishes a probability that a test spectrum matches a reference spectrum.
The problem being solved is increasing the selectivity of chemical sensors while reducing their complexity, size, and cost. Existing filter-based chemical sensors have challenges in discriminating chemicals with overlapping spectral signatures in complex backgrounds. Conventional methods, such as Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), reduce configuration-space dimensionality, which can limit discrimination. The CDSD approach overcomes these challenges by using multiple broadband infrared filters and expanding the dimensionality of the configuration space rather than reducing it, thereby enabling discrimination between chemicals with strongly overlapping IR spectra using simple optical filters and computational methods.
Unlike previous methods, CDSD uses low-resolution, large-bandwidth, overlapping spectral filters to construct chemical representative vectors and explores geometrical relationships between these vectors and surfaces or volumes they define in configuration space. This biomimetic approach relies on individual chemical responses to simple band-pass optical filters rather than complex optical elements, providing a novel data processing and detection approach for photometric systems. The described method is a non-iterative computational routine that differentiates chemicals even with high spectral overlap, allowing binary detection outcomes or relative density estimation of chemicals in mixtures.
Claims Coverage
The claims cover a comparative discrimination spectral detection system and method with several inventive features focusing on radiation delivery, collection, filtering, detection, and computational spectral analysis.
Geometrical shape-based spectral identification in configuration space
The system and method generate base vectors from reference spectra that define a polygon with volume in a configuration space, receive filtered test spectra to generate test vectors, assess their geometrical relationship to the polygon, and establish the probability of test spectrum matching a reference spectrum based on this assessment.
Use of low-resolution optical components with specific materials and configurations
The radiation source comprises an IR illuminator emitting in NIR, MWIR, or LWIR bands; radiation collector uses IR-compatible lenses and windows; beam splitters are optical plates with predetermined reflection and transmission; optical filters are glass filters with wavelength pass-bands peaked near wavelengths of interest; radiation detectors include thermal or photon detectors such as bolometers, pyroelectrics, thermopiles, Golay cells, photoconductors, photovoltaics, or photodiodes.
Geometrical volume inclusion criteria for spectral matching
If the test spectrum matches the reference spectrum, the corresponding test vector falls at least partially within the polygon volume in the configuration space and has predetermined relationships with the polygon's sides and surface normals; if not, the test vector falls at least partially outside the volume and lacks those relationships.
The claims define a system and method for identifying chemicals via spectral analysis, where spectral data is processed to form vectors in a configuration space that geometrically represent chemical signatures and allow for discrimination based on spatial inclusion within defined polygons, coupled with specified hardware components for radiation handling and detection.
Stated Advantages
The CDSD method provides high accuracy and low uncertainty in identifying chemicals even in the presence of highly similar overlapping spectral absorptions with a single spectrum collected in 10-20 seconds.
It enables discrimination between closely matched interferants and target chemicals with a computationally efficient, non-iterative algorithm that increases configuration-space dimensionality for analysis.
The system uses simple, low-resolution overlapping band-pass optical filters rather than complex optical elements, allowing for simpler computation and reduced system complexity, size, and cost.
The method supports binary (yes/no) or relative density detection outcomes and operates effectively in dynamically changing gaseous environments with target presence below 1 ppm detectable.
The computational operations are limited to finite matrix multiplications, facilitating implementation on small, handheld, and portable chemical sensing systems.
Documented Applications
Identification of chemicals with overlapping infrared spectral signatures using optical filter-based chemical sensors.
Discrimination of specific target chemicals in mixtures or spectrally cluttered backgrounds in near-infrared, mid-wave infrared, and long-wave infrared wavebands.
Rapid estimation and detection of chemical components in non-interacting chemical vapor mixtures, including vapors such as acetone, hexanes, and fuel oil.
Applications involving dynamically changing gaseous environments where chemical presence and concentration vary over time.
Interested in licensing this patent?