Video Block Decoding With Multiple Reference Line Prediction Modes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video coding methods rely on a single closest pixel line for intra prediction, which may include different textures or noise, leading to inefficiencies in selecting appropriate predictors and requiring excessive signaling of reference lines, and the encoder faces challenges in quickly choosing the best prediction mode from multiple options.
Innovation Solution
The proposed method allows for multiple-reference line prediction, where the encoder selects and signals multiple reference lines, enabling the decoder to determine the appropriate prediction mode based on predefined flags and indices, reducing the need for extensive signaling and improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple pixel lines are used to generate reference pixels and predictors, then prediction accuracy is improved, but signaling complexity increases due to requiring flags for selected reference lines
Solution Approach 1:
The patent segments the reference line selection process by introducing a flag-based mechanism that divides the multiple pixel lines into selectable categories. The encoder signals which reference lines to use through flags, allowing the decoder to efficiently select from pre-defined reference line candidates without processing all possible combinations, thus reducing signaling complexity while maintaining prediction accuracy.
Solution Approach 2:
The patent applies preliminary action by pre-defining mode lists for different reference lines before the actual prediction process. The encoder and decoder both have access to these pre-prepared mode lists, which contain prediction modes organized by reference line. This eliminates the need for the encoder to signal detailed mode selection information, as the decoder can independently select from the appropriate pre-defined list based on the signaled reference line flags.
2Measurement precision
If multiple reference lines are selected for prediction, then prediction accuracy is improved, but encoder processing time increases due to needing to select from multiple predicted results
Solution Approach 1:
The patent significantly reduces encoder processing time by pre-defining mode lists for different reference lines in advance. These mode lists are prepared once and stored, eliminating the need for the encoder to perform complex selection operations during encoding. The encoder only needs to signal simple flags indicating which pre-defined reference lines to use, while the decoder handles the more complex mode selection from the pre-prepared lists, thus transferring computational burden from encoder to decoder.
Solution Approach 2:
The patent enables the decoder to perform self-service by allowing it to independently select prediction modes from the pre-defined mode lists based on the reference line flags signaled by the encoder. This shifts the computational workload from the encoder to the decoder, reducing encoder processing time while maintaining prediction accuracy, as the decoder has full access to the pre-prepared mode information and can efficiently select the best mode without encoder intervention.
3Measurement precision
If different mode lists are used for different reference lines, then prediction accuracy is improved, but device complexity increases due to managing multiple mode lists
Solution Approach 1:
The patent applies universality by creating a unified framework where a single set of pre-defined mode lists serves multiple reference lines. Instead of requiring separate management mechanisms for each reference line's mode list, the system uses a universal mode list structure that can be applied to any reference line. The mode lists are organized in a standardized format that works across different reference lines, simplifying the overall system architecture and reducing device complexity while maintaining the ability to provide accurate predictions for each reference line type.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An aspect relates to a non-transitory machine-readable medium of an electronic device (12) storing one or more computer-executable instructions for decoding a bitstream, the one or more computer-executable instructions, when executed by at least one processing unit of the electronic device (12), causing the electronic device (12) to: determine a block unit having an intra prediction mode from an image frame according to the bitstream; determine, based on a specific reference line candidate and a most probable mode, mpm, flag of the block unit, the intra prediction mode of the block unit, wherein: a plurality of most probable modes is selected from a plurality of intra mode candidates, a plurality of remaining mode candidates is generated by removing the plurality of most probable modes from a plurality of intra mode candidates, the intra prediction mode is determined, without parsing the mpm flag, from the plurality of most probable modes when the specific reference line candidate is different from a first one of a plurality of reference line candidates; the intra prediction mode is determined from the plurality of most probable modes when the specific reference line candidate is identical to the first one of the plurality of reference line candidates and the mpm flag is equal to a predefined value, the intra prediction mode is determined from the plurality of remaining mode candidates when the specific reference line candidate is identical to the first one of the plurality of reference line candidates and the mpm flag is different from the predefined value, and the first one of the plurality of reference line candidates is immediately adjacent to the block unit; determine the intra prediction mode of the block unit based on a first one of a plurality of mode indices, when the intra prediction mode is selected from the most probable modes, wherein the first one of the plurality of mode indices is an mpm index; determine the intra prediction mode of the block unit based on a second one of the plurality of mode indices, when the intra prediction mode is selected from the remaining mode candidates; and reconstruct the block unit of the image frame based on the intra prediction mode.