Video Coding Weighted Prediction Signaling Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution, high-quality images/videos, particularly in fields like VR/AR and Ultra High Definition, necessitates a highly efficient compression technique to reduce transmission and storage costs, while existing methods struggle with efficient inter prediction and signaling overhead.
Innovation Solution
A method and apparatus for enhancing video/image compression efficiency through weighted prediction, where a flag related to weighted prediction is parsed from a bitstream, and a prediction weighted table syntax is generated to perform efficient inter prediction, reducing signaling overhead by including this syntax in a picture or slice header.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high resolution and high quality image/video are transmitted or stored using existing methods, then image quality is maintained, but transmission and storage costs increase significantly
Solution Approach 1:
The patent extracts and transmits only the essential difference information (residuals) between original and predicted blocks rather than transmitting complete high-resolution blocks. By using inter prediction to extract motion information and weighted prediction to capture gradient variations, the method transmits only the necessary correction data, dramatically reducing transmission and storage requirements while maintaining image quality.
Solution Approach 2:
The patent changes the representation parameters by using weighted prediction with gradient-based weight calculations instead of uniform prediction. This parameter change allows more efficient encoding of high-frequency variations in high-resolution images, reducing the number of bits needed to represent the same visual information.
2Device complexity
If conventional inter prediction methods are used, then encoding simplicity is maintained, but prediction accuracy for high-resolution content deteriorates
Solution Approach 1:
The patent performs preliminary weighted prediction using gradient information before final reconstruction. By pre-calculating weights based on gradient magnitudes and applying them to prediction samples, the method prepares enhanced prediction data in advance, improving prediction accuracy for high-resolution content without significantly increasing overall encoding complexity.
Solution Approach 2:
The patent introduces gradient information as an intermediary element between motion compensation and final prediction. The gradient-based weights act as a mediator that refines the basic inter prediction output, capturing local variations and edges in high-resolution blocks without requiring complex direct modeling.
3Measurement precision
If weighted prediction information is signaled for every block, then prediction accuracy is improved, but signaling overhead increases
Solution Approach 1:
The patent segments the weighting information signaling by introducing a block-type flag that divides blocks into different categories (e.g., homogeneous vs. heterogeneous regions). This segmentation allows weighted prediction to be applied selectively only where needed, reducing signaling overhead while maintaining prediction accuracy for blocks that benefit from it.
Solution Approach 2:
The patent applies weighted prediction partially rather than universally by using the block-type flag to determine when to apply the technique. This partial action approach applies the computationally intensive weighted prediction only to blocks where it provides significant benefit, reducing overall signaling overhead while maintaining sufficient prediction accuracy for the application.
Data Source
AI summary
A video decoding method performed by a video decoding apparatus, according to the present document, comprises the steps of: parsing a flag for weighted prediction from a bitstream; parsing a prediction weighted table syntax from the bitstream on the basis of the flag; generating prediction samples by performing weighted prediction on a current block within a current picture on the basis of the prediction weighted table syntax; and generating reconstructed samples for the current block on the basis of the prediction samples, wherein the prediction weighted table syntax may be parsed from a picture header or slice header of the bitstream on the basis of a value of the flag.