Optical 4f system performing extended convolution operation and operating method thereof
AssigneeElectronics and Telecommunications Research Institute
InventorsIk-Soon Kim, Jin Hwa Gene
Life of this patent
Estimating the term…
Abstract
Disclosed is an optical system which includes a first lens that receives input data from an object, a kernel that performs a first Fourier transform on the input data and generates learning data by performing calculation on a result of the first Fourier transform and pattern data, and a second lens that generates result data by performing a second Fourier transform on the learning data, and the input data, the learning data, and the result data include both a positive number and a negative number.
Claims16
An optical system comprising:
- a first lens configured to receive input data from an object and to perform a first Fourier transform of the input data such that a Fourier-transformed signal is formed at a location of a kernel;
- the kernel disposed at the location at which the Fourier-transformed signal is formed, the kernel being configured to generate learning data by performing a convolution operation on a result of the first Fourier transform of the input data and pattern data;
- and
- a second lens configured to generate result data by performing a second Fourier transform on the learning data,
- wherein values used in the convolution operation of the optical system are defined based on a light amplitude, while actual input values and actual output values of the optical system are proportional to a light intensity,
- wherein the input data are representable as a difference between a first non-negative input signal and a second non-negative input signal,
- wherein the pattern data are representable as a difference between a first non-negative pattern signal and a second non-negative pattern signal, and
- wherein the result data include both a positive number and a negative number and are obtained by:
- performing optical calculations on non-negative signals corresponding to the input data and the pattern data without generating a negative light intensity, and
- recombining results of the optical calculations to generate the result data including both the positive number and the negative number.
An operating method of an optical system which includes a first lens, a kernel, and a second lens, the operating method comprising:
- receiving input data from an object through the first lens;
- performing a first Fourier transform on the input data such that a Fourier-transformed signal is formed at a location of the kernel;
- generating learning data by performing a convolution operation on a result of the first Fourier transform of the input data and pattern data;
- and
- generating result data by performing a second Fourier transform on the learning data,
- wherein values used in the convolution operation of the optical system are defined based on a light amplitude, while actual input values and actual output values of the optical system are proportional to a light intensity, and
- wherein the result data include both a positive number and a negative number, the result data being obtained by:
- representing the input data as a difference between a first non-negative input signal and a second non-negative input signal,
- representing the pattern data as a difference between a first non-negative pattern signal and a second non-negative pattern signal,
- performing optical calculations on non-negative signals corresponding to the input data and the pattern data without generating a negative light intensity, and
- recombining results of the optical calculations to generate the result data including both the positive number and the negative number.
Dependencies are inferred from the claim text.