Sample-Based Angular Intra-Prediction for Lossless Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards, such as HEVC, are inherently lossy, which is undesirable in applications like automotive vision and web collaboration where lossless coding is required.
Innovation Solution
The implementation of sample-based angular intra-prediction (SAP) in video coding, which generates predicted samples by linear interpolation of adjacent reference samples, allowing for lossless coding when enabled.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If hybrid coding scheme using block-based prediction and transform coding is used, then compression efficiency is improved, but lossless coding capability deteriorates
Solution Approach 1:
The prediction unit is divided into multiple sub-regions, with different prediction methods applied to different sub-regions. Specifically, the PU is divided into a first sub-region and a second sub-region, where the first sub-region uses sample-based angular intra-prediction and the second sub-region uses block-based angular intra-prediction, allowing both lossless and lossy coding capabilities to coexist within the same prediction unit
Solution Approach 2:
Different prediction modes are applied to different spatial regions within the prediction unit based on local characteristics. The sample-based angular intra-prediction is applied to regions requiring lossless coding while block-based angular intra-prediction is applied to other regions, optimizing both lossless capability and compression efficiency locally
2Productivity
If quantization of transform coefficients is applied, then compression ratio is improved, but video quality deteriorates
Solution Approach 1:
The quantization step is extracted and selectively applied only to specific regions where lossy coding is acceptable. By using sample-based angular intra-prediction without quantization in regions requiring high quality, the patent removes the harmful quantization effect from critical areas while maintaining compression efficiency in less critical areas
Data Source
AI summary
A method for processing a prediction unit (PU) to generate predicted samples is provided that includes computing predicted samples for samples of the PU using sample-based angular intra-prediction (SAP) when lossless coding is enabled for the PU, and computing predicted samples for the samples of the PU using block-based angular intra-prediction when lossless coding is not enabled for the PU. Computation of the predicted using SAP includes determining an intra-prediction angle for the PU, and computing a predicted sample for each sample of the samples in the PU based on linear interpolation of two reference samples adjacent to the sample, wherein the two reference samples are selected according to the intra-prediction angle.