MRL Intra Prediction With DC-Mode MPM Lists for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image/video coding technologies face inefficiencies in compressing high-resolution, high-quality images/videos, particularly for applications like virtual reality and augmented reality, leading to increased transmission and storage costs.
Innovation Solution
Implementing Multi-Reference Line (MRL)-based intra prediction with a Most Probable Mode (MPM) list that includes a DC mode, enhancing prediction accuracy and reducing computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If high resolution, high quality image/video is transmitted or stored, then image quality is improved, but transmission and storage costs are increased
Solution Approach 1:
The patent applies parameter changes by introducing Multi-Reference Line (MRL) based intra prediction with multiple reference lines (first reference line, second reference line, third reference line) at different positions. This changes the prediction parameters from conventional single reference line to multiple reference lines, enabling more accurate prediction for high resolution images while reducing bit rate requirements, thus lowering transmission and storage costs.
Solution Approach 2:
The patent segments the prediction process into multiple stages by dividing the prediction block into first, second, and third prediction blocks corresponding to different reference lines. Each segment uses different reference lines (first reference line for first prediction block, second reference line for second prediction block, third reference line for third prediction block), allowing independent optimization of prediction accuracy for different regions while reducing overall data requirements.
2Measurement precision
If conventional intra prediction is used, then device complexity is low, but prediction accuracy is insufficient for high resolution images
Solution Approach 1:
The patent segments the prediction mechanism into multiple independent parts: first prediction block with first reference line, second prediction block with second reference line, and third prediction block with third reference line. Each segment operates independently with its own reference line, allowing the system to achieve high prediction accuracy through multiple simplified prediction operations rather than one complex operation.
Solution Approach 2:
The patent adds a new dimension to the prediction mechanism by introducing multiple reference lines at different positions (first reference line, second reference line, third reference line) instead of using a single reference line. This dimensional expansion from one reference line to multiple reference lines enables the system to capture more spatial information and improve prediction accuracy for high resolution images.
3Reliability
If DC mode is not included in MPM list for MRL-based intra prediction, then MPM list construction is simple, but prediction performance is degraded
Solution Approach 1:
The patent merges the DC mode with the MRL-based intra prediction by including DC mode in the MPM list. The decoding apparatus constructs the MPM list by deriving candidate intra prediction modes and explicitly includes DC mode as one of the candidates. This merging allows the system to leverage both DC mode and MRL-based prediction for comprehensive prediction coverage, improving prediction performance by providing multiple prediction options.
Solution Approach 2:
The patent applies preliminary action by pre-constructing the MPM list with candidate intra prediction modes including DC mode before the actual prediction process. The MPM list is constructed in advance with multiple candidate modes (including DC mode) so that the decoding apparatus can efficiently select the best prediction mode during decoding without needing to construct the list dynamically, thus improving prediction performance while keeping the construction process manageable.
Data Source
AI summary
An image decoding method includes: configuring a MPM list including candidate intra prediction modes for a current block; deriving an intra prediction mode of the current block from the MPM list on the basis of MPM index information indicating an intra prediction mode, among the candidate intra prediction modes included in the MPM list, for the current block; generating prediction samples for the current block on the basis of the intra prediction mode; and generating a reconstructed picture for the current block on the basis of the prediction samples, wherein the step for configuring the MPM list comprises, on the basis of the case where the value of reference line index information, representing a reference line used for the intra prediction of the current block, is not zero, deriving a DC mode as one mode among the candidate intra prediction modes, and including same in the MPM list.