Weighted Prediction Parameter Calculation for Image Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image coding methods face inefficiencies in compression due to luminosity variations between images, particularly when using weighted prediction methods, which either incur high bitrate costs by transmitting parameters or lose precision by not accounting for local content variations.
Innovation Solution
A method for encoding and decoding that dynamically adjusts weighted prediction parameters based on local content and illumination variations, without explicitly transmitting these parameters, by calculating them on both the coder and decoder sides, and using template matching for motion data derivation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If weighted prediction parameters are explicitly transmitted in the stream per image slice, then prediction accuracy is improved, but bitrate overhead increases
Solution Approach 1:
The decoder calculates weighted prediction parameters autonomously using only received coded data and stored reference images, without requiring explicit parameter transmission. The decoder computes the ratio between the average of the current block and the average of the reference block, serving itself by deriving parameters from the data already available in the system.
Solution Approach 2:
The invention extracts and utilizes only the essential information (block averages and reference block averages) that is already present in the coded data and reference images, eliminating the need to transmit separate weighted prediction parameters. This extraction approach removes the harmful element (parameter transmission overhead) while preserving the useful function (prediction accuracy).
2Quantity of substance
If weighted prediction parameters are calculated locally without transmission, then bitrate overhead is reduced, but prediction precision is lost
Solution Approach 1:
The decoder performs self-service by calculating weighted prediction parameters locally using the same method as the encoder (ratio of block averages), ensuring both sides arrive at identical parameter values without transmission. This maintains prediction precision while avoiding bitrate overhead.
Solution Approach 2:
The invention changes the approach from transmitting parameter values to calculating parameter values locally using a specific formula (ratio of averages). This parameter change in the calculation method enables both low overhead and high precision simultaneously.
3Productivity
If inter-image prediction is used, then compression efficiency is improved, but performance deteriorates when luminosity variation occurs
Solution Approach 1:
The invention dynamically adapts the prediction method based on luminosity variation detection. When luminosity variation is detected, the system switches to using weighted prediction parameters calculated from block averages, allowing the prediction mechanism to adjust its behavior dynamically to maintain reliability under changing conditions.
Solution Approach 2:
The system changes the prediction parameter calculation method in response to luminosity variation. By computing weighted prediction parameters as the ratio of current block average to reference block average, the system adapts to luminosity changes and maintains reliable prediction while preserving compression efficiency.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for decoding a stream of coded data (F) representative of a sequence of images comprising the following steps for a current block: reconstructing (10) a residual block, determining (12) a prediction block, and reconstructing (14) the current block by merging the residual block and the prediction block weighted by a weighted prediction parameter. According to the invention, the weighted prediction parameter is determined as follows: calculating (2) the average of the prediction block (avrpred), reconstructing (4) a DC coefficient of the current block (IresDC), calculating (6) an average (avrcr_dec) of the current block from the average of the prediction block (avrpred) and the DC coefficient (IresDC), and calculating (8) the weighted prediction parameter being the ratio of the average of the current block and the average of the prediction block (avrpred). The invention also relates to a method for coding.