Super-Resolution Video Coding With Low-Resolution Motion Compensation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding technologies face challenges in efficiently managing memory consumption and computational complexity, particularly in video compression systems, due to the need for motion vector coding and motion compensation between pictures with different spatial resolutions, and the inefficiencies in coding residual information.

Innovation Solution

Incorporating a neural network approach for super-resolution restoration with residual frame coding, allowing for encoding and decoding operations to be performed at different spatial resolutions, and utilizing a multi-scale machine learning model to enhance video coding efficiency and reduce computational complexity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If video coding is performed at high spatial resolution, then video quality is improved, but memory consumption and computational complexity increase

Engineering Contradiction:
Improvevideo qualityVSAvoidcomputational complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The video coding process is segmented into two distinct stages: low-resolution motion compensation and high-resolution residual coding. This segmentation allows the computationally intensive motion estimation to be performed at low resolution while maintaining final output quality at high resolution, thereby reducing overall computational complexity without sacrificing video quality

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a resolution dimension transition by performing motion compensation at low resolution and then refining the residual at high resolution. This dimensional transition in processing resolution allows the system to avoid the computational burden of full high-resolution motion estimation while still achieving high-quality output

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Adaptability or versatility

If motion compensation is performed between pictures with different spatial resolutions, then coding flexibility is improved, but memory consumption increases

Engineering Contradiction:
Improvecoding flexibilityVSAvoidmemory consumption
Core Design Contradiction:
Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

Different resolution qualities are applied to different stages of the coding process: low resolution for motion compensation operations and high resolution for residual coding. This local quality differentiation allows the system to use only the necessary resolution for each specific operation, reducing overall memory consumption while maintaining coding flexibility

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent implements dynamic resolution switching where the processing resolution changes based on the coding stage. The system dynamically transitions from low-resolution processing during motion compensation to high-resolution processing during residual coding, optimizing memory usage at each stage rather than maintaining high resolution throughout

Inventive Principle:
Principle #15Dynamics

3Measurement precision

If residual information is coded at high resolution, then coding accuracy is improved, but computational complexity increases

Engineering Contradiction:
Improvecoding accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The coding process is segmented such that only the residual information (the difference between predicted and actual values) is processed at high resolution, while the bulk motion compensation operations occur at low resolution. This segmentation ensures high coding accuracy for the critical residual portion without applying high-resolution processing to all data

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS12375697B2Computer-implemented method and apparatus for video coding using super-resolution restoration with residual frame coding
Publication Date: 2025.07.29 AMAZON TECH INC
  • US12375697B2 patent drawing
  • US12375697B2 patent drawing
  • US12375697B2 patent drawing

AI summary

The present disclosure relates to methods, apparatus, systems, and non-transitory computer-readable storage media for video coding using super-resolution restoration with residual frame coding. According to some examples, a computer-implemented method includes receiving a coded frame of a video; performing a video coding on the coded frame of the video to generate a resultant for the coded frame at a second lower resolution than a first resolution; upsampling the resultant in at least a vertical direction to a higher resolution than the second lower resolution to generate an upsampled resultant; generating a decoded frame based on at least the upsampled resultant; and transmitting the decoded frame to a frame buffer or to a display device.