Video Block Prediction-Mode Signaling with Derived Flags
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding methods require a large number of bits to represent prediction mode information, leading to high encoding costs and poor decoding efficiency.
Innovation Solution
The method involves determining prediction-mode flag information for a current block, where flag values are derived based on decoding conditions, reducing the need for explicit flag values in the encoded bitstream, thereby minimizing the number of bits required and simplifying the decoding process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If indication information of the target prediction mode is carried in the encoded bitstream, then the decoder can determine the target prediction mode, but a large number of bits are used leading to high encoding cost and low decoding efficiency
Solution Approach 1:
The patent extracts only the necessary prediction mode information from the bitstream by using context-based derivation. Instead of transmitting complete mode indicators, the system transmits minimal syntax elements (flags) that allow the decoder to derive the target prediction mode through predefined rules and context information, thereby reducing the number of bits while maintaining decoding accuracy
Solution Approach 2:
The patent inverts the traditional approach by having the decoder derive the prediction mode from minimal flags rather than having the encoder transmit the complete mode information. The derivation process uses context information and predefined rules to reconstruct the target prediction mode, achieving compression without loss of decoding capability
2Manufacturing precision
If multiple prediction modes are supported for video encoding, then coding performance is improved, but the complexity of determining the target prediction mode increases
Solution Approach 1:
The patent segments the prediction mode selection process into multiple stages: first determining whether to use merge mode via a flag, then conditionally deriving the target prediction mode based on block characteristics and context. This segmentation allows multiple prediction modes to be supported while managing complexity through hierarchical decision-making
Solution Approach 2:
The patent applies different derivation rules and context information based on local block characteristics such as block size, position, and neighboring block availability. Each block's target prediction mode is determined using locally optimized rules rather than a universal complex algorithm, reducing overall decoding complexity while maintaining coding performance
Data Source
Figure 1~3
Figure 4A~4B
Figure 4C~4D
AI summary
The present disclosure provides a decoding method and apparatus, an encoding method and apparatus, and a device. The decoding method includes: determining prediction-mode flag information of a current block, the prediction-mode flag information includes a flag value of each of one or more of N candidate prediction modes; when the flag value is a first value, the flag value indicates that a candidate prediction mode is enabled, and when the flag value is a second value, the flag value indicates that the candidate prediction mode is disabled, and for the flag value of one of the N candidate prediction modes, if the current block does not satisfy a decoding condition of the flag value of the candidate prediction mode, the flag value is a derived flag value of the candidate prediction mode; determining a target prediction mode used by the current block according to the prediction-mode flag information, where the target prediction mode is one of the N candidate prediction modes; and decoding the current block according to the target prediction mode. The present disclosure can improve the encoding performance.