Chrominance Subsampling Phase Control for Video Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video conferencing compression techniques, such as chrominance subsampling, often result in image degradation and color artifacts for content with sharp edges, like graphics and text, due to the loss of spectral information during encoding and decoding, especially when using 4:2:0 data formats instead of 4:4:4.
Innovation Solution
A method and device for video conferencing that determines a sampling phase for chrominance subsampling and up-sampling, synthesizing motion to transmit additional spectral information, maintaining image quality similar to 4:4:4 sampling while using existing pipelines, and ensuring backward compatibility by processing devices with different hardware and software components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If chrominance subsampling is used to reduce data bandwidth, then transmission efficiency is improved, but image quality deteriorates due to loss of spectral information
Solution Approach 1:
The patent applies preliminary action by performing chrominance subsampling at multiple different phase positions (e.g., quarter-phase shifts) before encoding. This allows the system to capture spectral information at different sampling points, and then reconstruct the full spectral content during decoding by combining information from multiple phases, thereby maintaining image quality while still achieving bandwidth reduction
Solution Approach 2:
The patent changes the sampling phase parameter during the subsampling process. By varying the phase position at which chrominance subsampling is performed (e.g., using phases offset by 1/4 of the sampling period), the system captures different spectral information from the same signal, enabling reconstruction of lost spectral components and improving reconstructed image quality
2Quantity of substance
If chrominance subsampling is applied to compress video signals, then data bandwidth is reduced, but color accuracy deteriorates causing color bleeding artifacts
Solution Approach 1:
The system performs preliminary chrominance subsampling at multiple phase positions before encoding. By capturing chrominance information at different phase offsets (e.g., 0, 1/4, 1/2, 3/4 of the sampling period), the system preserves color information that would otherwise be lost, enabling accurate color reconstruction and preventing color bleeding artifacts
Solution Approach 2:
The patent combines chrominance samples from multiple different phase positions to create a composite representation of the original color information. This composite approach allows the system to reconstruct full-color accuracy from subsampled data by integrating information from different sampling phases, effectively preventing color bleeding while maintaining reduced bandwidth
3Productivity
If 4:2:0 data format is used instead of 4:4:4, then transmission efficiency is improved, but sharpness of edges deteriorates
Solution Approach 1:
The patent performs preliminary chrominance subsampling at multiple phase positions (including quarter-phase shifts) before encoding in 4:2:0 format. This allows the system to capture edge information at different sampling points, and then reconstruct sharp edges during decoding by combining the multi-phase information, maintaining edge sharpness comparable to 4:4:4 while achieving 4:2:0 bandwidth efficiency
Data Source
AI summary
A processing device for use with a video conferencing network is provided. The processing device includes memory configured to store data and a processor. The processor is configured to determine a first sampling phase for a portion of first video data and chrominance sub-sample the portion of first video data using the first sampling phase. The processor is also configured to encode the sub-sampled portion of first video data and decode a sub-sampled, encoded portion of second video data. The processor is further configured to determine a second sampling phase at which the portion of second video data is chrominance sub-sampled and chrominance up-sample the portion of second video data using the second sample phase.