Apparatus and method for reconstructing image using extended neural network
AssigneeKorea Advanced Institute of Science and Technology
InventorsJongChul Ye, YoSeob Han, Eunju Cha
Life of this patent
Estimating the term…
Abstract
A method and apparatus for reconstructing an image using an extended neural network is provided. The method includes receiving an input image and reconstructing an output image from the input image using a neural network meeting a predetermined frame constraint. The reconstructing includes transforming the input image into signals corresponding to different frequencies, adjusting coefficients of the transformed signals using a nonlinear function, reconstructing the adjusted coefficients, and inversely transforming all coefficients determined using the reconstructed coefficients into the output image.
Claims19
A method for reconstructing an image, the method comprising:
- receiving an input image;
- and
- reconstructing an output image from the input image using a neural network meeting a predetermined frame constraint, the reconstructing the output image comprising:
- transforming the input image into signals corresponding to different frequencies;
- adjusting coefficients of the transformed signals using a nonlinear function;
- reconstructing the adjusted coefficients;
- and
- inversely transforming all coefficients determined using the reconstructed coefficients into the output image.
A method for reconstructing an image, the method comprising:
- transforming an input image into signals corresponding to different frequencies;
- adjusting coefficients of the transformed signals using a nonlinear function;
- reconstructing the adjusted coefficients using a previously learned neural network;
- and
- inversely transforming all coefficients determined using the reconstructed coefficients into an output image.
An apparatus for reconstructing an image, the apparatus comprising:
- non-transitory memory storing computer-readable instructions;
- and
- at least one processor,
- wherein the at least one processor is configured to execute the computer-readable instructions to:
- receive an input image;
- and
- reconstruct an output image from the input image using a neural network meeting a predetermined frame constraint, the at least one processor reconstructing the output image by being configured to execute the computer-readable instructions to:
- transform the input image into signals corresponding to different frequencies;
- adjust coefficients of the transformed signals using a nonlinear function;
- reconstruct the adjusted coefficients;
- and
- inversely transform all coefficients determined using the reconstructed coefficients into the output image.
Dependencies are inferred from the claim text.