Integrated photonic encoding systems and methods for high speed and low power image processing
AssigneesNational Technology & Engineering Solutions of Sandia, LLC, U.S. Naval Research Laboratory, Arizona Board of Regents on Behalf of the University of Arizona, Tech Launch Arizona
InventorsRaktim Sarma, Jayson Briscoe, Michael Goldflam, Nicholas Karl, Christopher Long, Brandon Redding, David Brady
Life of this patent
Estimating the term…
Abstract
A method performed by a photonics processing system includes encoding an input vector into first optical signals; performing an optical scattering of the first optical signals to form second optical signals, the optical scattering implementing a matrix multiplication of the input vector by a quasi-random matrix; detecting at least a portion of the second optical signals representing an output vector; and encoding the second optical signals in an electrical representation of the output vector.
Claims13
A photonic processing system, comprising:
- an optical encoder;
- a photonic processor in optical communication with the optical encoder;
- and
- an optical receiver in optical communication with the photonic processor;
- wherein the optical encoder is configured to encode an input vector into first optical signals;
- wherein the photonic processor includes a scattering region having scattering centers, wherein a refractive index of the scattering centers is different than a refractive index of a material of a remainder of the scattering region and is configured to:
- receive the first optical signals;
- perform an optical scattering on the first optical signals to form second optical signals, the optical scattering implementing a matrix multiplication of the input vector by a quasi-random matrix;
- and
- output the second optical signals representing an output vector;
- and
- wherein the optical receiver is configured to detect at least a portion of the second optical signals and output an electrical representation of the output vector;
- and
- wherein the scattering centers are cylindrical voids extending through a thickness of the photonic processor, wherein the cylindrical voids have a refractive index of about 1, and wherein the refractive index of the material comprising the photonic processor is greater than 1.
A photonic processing system, comprising:
- an optical encoder;
- a photonic processor in optical communication with the optical encoder;
- and
- an optical receiver in optical communication with the photonic processor;
- wherein the optical encoder is configured to encode an input vector into first optical signals;
- an output port having spatial regions;
- wherein the photonic processor is configured to:
- receive the first optical signals;
- perform an optical scattering on the first optical signals to form second optical signals, the optical scattering implementing a matrix multiplication of the input vector by a quasi-random matrix;
- and
- output the second optical signals representing an output vector;
- and
- wherein the optical receiver is configured to detect at least a portion of the second optical signals and output an electrical representation of the output vector,
- wherein:
- the quantity of optical receivers corresponds to at least one of a size of the output vector or the quantity of spatial regions;
- the quantity of optical signal modulators corresponds to at least one of a size of the input vector or a size of the quasi-random matrix;
- and
- a ratio of the quantity of the optical signal modulators to the quantity of the optical receivers corresponds to a compression ratio;
- and
- wherein the portion of the second optical signals corresponds to one of the spatial regions.
A method for photonically processing data, the method comprising:
- encoding an input vector into first optical signals;
- performing an optical scattering of the first optical signals to form second optical signals, the optical scattering implementing a matrix multiplication of the input vector by a quasi-random matrix;
- detecting at least a portion of the second optical signals representing an output vector;
- and
- encoding the second optical signals in an electrical representation of the output vector;
- encoding a second input vector into third optical signals;
- performing an optical scattering of the third optical signals to form fourth optical signals, the optical scattering implementing a matrix multiplication of the second input vector by the quasi-random matrix;
- detecting the fourth optical signals representing a second output vector;
- and
- encoding the fourth optical signals in an electrical representation of the second output vector;
- wherein the second input vector corresponds to second pixels comprising a second pixel block of the source image;
- wherein the pixel block and the second pixel block are non-overlapping;
- wherein:
- elements of the input vector correspond to pixels in a source image,
- the pixels comprise a pixel block of the source image, and
- the source image comprises pixel blocks.
Dependencies are inferred from the claim text.