Reference Sample Position Derivation for Adaptive Resolution Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards lack the ability to adaptively change resolution without requiring instantaneous decoding refresh (IDR) or intra random access points, which can lead to inefficiencies and poor user experience in scenarios like video conferencing and streaming, especially when network conditions change or active speakers are adjusted.
Innovation Solution
Implement methods for video processing that include adaptive resolution change (ARC) and reference picture resampling (RPR) to handle resolution changes seamlessly, using techniques such as fractional sample interpolation and conformance windows to manage motion vectors and enable efficient conversion between different resolutions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If adaptive resolution change is implemented without IDR or intra random access points, then resolution adaptation efficiency is improved, but video coding complexity increases
Solution Approach 1:
The patent changes the parameter of resolution by using reference picture resampling with different sampling rates. The encoder can resample reference pictures at different resolutions (e.g., from 4K to 1080p) by adjusting the sampling rate, enabling adaptive resolution change without requiring full IDR or intra random access points, thus improving resolution adaptation efficiency while managing coding complexity
Solution Approach 2:
The patent performs preliminary resampling of reference pictures before motion compensation. By pre-processing reference pictures at the desired resolution, the system avoids the need for complex real-time resolution switching during decoding, reducing the computational burden on the decoder and managing overall video coding complexity
2Loss of time
If reference picture resampling is used for resolution adaptation, then latency is reduced, but processing complexity increases
Solution Approach 1:
The patent replaces traditional mechanical resolution switching mechanisms (which require full picture re-encoding and IDR points) with a signal processing approach using resampling filters. This substitution allows for faster, lower-latency resolution adaptation by simply filtering and downsampling reference pictures, significantly reducing the time required for resolution changes while managing processing complexity through efficient filter design
3Measurement precision
If fractional sample interpolation is applied, then motion vector precision is improved, but computational complexity increases
Solution Approach 1:
The patent applies fractional sample interpolation selectively rather than universally. By using interpolation only where needed (e.g., for sub-pixel motion compensation in specific blocks) and using integer-sample references elsewhere, the system achieves improved motion vector precision where necessary while avoiding the computational complexity of applying fractional interpolation to every sample in every block
Data Source
AI summary
An example method of video processing includes determining, for a conversion between a current picture of a video and a coded representation of the video, a position of a reference sample in a reference picture that is associated with the current picture based on a top-left position of a window of a picture. The picture includes at least the current picture or the reference picture, and the window is subject to a processing rule during the conversion. The method also includes performing the conversion based on the determining.