Method and apparatus for video coding using an improved in-loop filter
AssigneesHyundai Motor Company, Kia Corporation, Ewha University-Industry Collaboration Foundation
InventorsJe Won Kang, Jung Kyung Lee, Seung Wook Park, Jin Heo
Life of this patent
Estimating the term…
Abstract
A method and an apparatus are disclosed for video coding using an improved in-loop filter. The video coding method and the apparatus generate a residual frame from a reconstructed frame using a deep learning model. The video coding method and the apparatus improve performance of an in-loop filter by approximating an original residual frame by applying the generated residual frame to a linear model.
Claims19
A method for improving video quality of a reconstructed frame, performed by a video encoding apparatus, the method comprising:
- obtaining the reconstructed frame and inputting the reconstructed frame into a deep learning-based improvement model to generate an output that approximates an original frame, wherein the reconstructed frame is a frame obtained by reconstructing the original frame and reconstructed in advance by the video encoding apparatus;
- generating a first residual frame by subtracting the reconstructed frame from an output of the deep learning-based improvement model;
- inputting the first residual frame into a linear model to generate a second residual frame, wherein the linear model includes parameters representing a linear relation between the first residual frame and the second residual frame;
- and
- generating an improved reconstructed frame by adding the second residual frame and the reconstructed frame.
A method for improving video quality of a reconstructed frame, performed by a video decoding apparatus, the method comprising:
- obtaining the reconstructed frame and inputting the reconstructed frame into a deep learning-based improvement model to generate an output that approximates an original frame, wherein the reconstructed frame is a frame obtained by reconstructing the original frame and reconstructed in advance by the video decoding apparatus;
- generating a first residual frame by subtracting the reconstructed frame from an output of the deep learning-based improvement model;
- inputting the first residual frame into a linear model to generate a second residual frame, wherein the linear model includes parameters representing a linear relation between the first residual frame and the second residual frame;
- and
- generating an improved reconstructed frame by adding the second residual frame and the reconstructed frame.
A computer-readable recording medium storing a bitstream generated by a video encoding method for improving video quality of a reconstructed frame, the video encoding method comprising:
- obtaining the reconstructed frame and inputting the reconstructed frame into a deep learning-based improvement model to generate an output that approximates an original frame, wherein the reconstructed frame is a frame obtained by reconstructing the original frame and reconstructed in advance;
- generating a first residual frame by subtracting the reconstructed frame from an output of the deep learning-based improvement model;
- inputting the first residual frame into a linear model to generate a second residual frame, wherein the linear model includes parameters representing a linear relation between the first residual frame and the second residual frame;
- and
- generating an improved reconstructed frame by adding the second residual frame and the reconstructed frame.
Dependencies are inferred from the claim text.