Multi-Hypothesis Prediction Signaling for Lower-Bit Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing multi-hypothesis prediction (MHP) design in video coding is inefficient due to redundant signaling of the MHP merge flag and suboptimal motion information list construction, leading to wasteful bit usage and poor prediction accuracy.
Innovation Solution
Implement conditional signaling of the MHP merge flag and enhance the motion information list construction process by considering both L0 and L1 motion vectors from previously decoded blocks and utilizing the current block's motion information for better fit, along with a refinement process to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the MHP merge flag is always signaled in the bitstream, then the decoder can correctly interpret motion information, but redundant bit usage occurs when the flag is constant
Solution Approach 1:
The patent makes the MHP merge flag signaling dynamic by conditionally including it only when its value differs from a default value. The flag is present in the bitstream only when necessary (when mhpMergeFlag ≠ 0), and omitted when redundant (when mhpMergeFlag = 0), allowing the decoding process to adapt to each specific case rather than following a fixed signaling pattern.
Solution Approach 2:
The patent extracts and removes the redundant MHP merge flag signaling from the bitstream. By checking whether the flag value differs from the default (0), the patent selectively excludes unnecessary flag transmissions, keeping only the essential information needed for correct decoding while eliminating wasteful bit usage.
2Device complexity
If the motion information list is constructed using only one L0 or L1 motion vector from previously decoded blocks, then the construction process is simple, but prediction accuracy deteriorates
Solution Approach 1:
The patent merges multiple motion information sources into a unified candidate list. It combines L0 and L1 motion vectors from the first previously decoded block with motion vectors from the second previously decoded block, creating a comprehensive list of motion information candidates that leverages the strengths of multiple sources for improved prediction accuracy.
Solution Approach 2:
The patent creates a universal motion information list construction process that can handle multiple types of motion vectors (L0, L1) from multiple blocks. The constructed list serves multiple functions: it provides candidates for both implicit and explicit signaling modes, and works for different block types and prediction scenarios, making the system more versatile and adaptable.
3Loss of substance
If implicit signaling is used for the second piece of motion information, then bit usage is reduced, but the decoder cannot directly determine the signaling mode
Solution Approach 1:
The patent introduces a feedback mechanism where the decoder determines the signaling mode based on the presence or absence of the second syntax element. When the element is absent, the decoder infers implicit signaling; when present, it decodes the value to determine the mode. This feedback loop allows the system to use minimal bits while maintaining unambiguous mode identification through the structure of the bitstream itself.
Data Source
AI summary
Methods and apparatuses for processing a current block within a current picture. The processing may include determining whether a second syntax element S1 is present in a coded video bitstream. The processing may include, if the S1 is not present, inferring which one of implicit signaling and explicit signaling is used to signal a second piece of motion information M1, or, if the S1 is present, determining which one of implicit signaling and explicit signaling is used to signal the M1 based on a decoded value of the S1. The processing may include, if implicit signaling is inferred or determined, deriving the M1, or, if explicit signaling is inferred or determined, determining the M1. The processing may include generating a prediction block Pb01 for the current block based on a first piece of motion information M0, the M1, and a weight factor W1. The second syntax element S1 may be a multi-hypothesis prediction, MHP, merge flag and the third syntax element S2 may be a motion information index.