Video Decoder Weight Prediction for Brightness Change Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding/decoding methods face inefficiencies in inter prediction, particularly when there are brightness changes between images or multiple light sources, leading to inaccurate predictions and increased residual energy.
Innovation Solution
The method determines weight prediction parameters based on brightness changes between images, using a combination of first and second weight prediction parameters to adaptively improve inter prediction efficiency, even in cases of local brightness changes or multiple light sources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If inter prediction is performed using conventional methods without weight prediction, then the encoding/decoding process is simple, but prediction accuracy deteriorates when brightness changes occur between images
Solution Approach 1:
The patent applies parameter changes by introducing weight prediction parameters (multiplication and addition parameters) that dynamically adjust the prediction process. These parameters are derived based on brightness change detection between current and reference images, allowing the system to adapt to varying brightness conditions and improve prediction accuracy without requiring complete redesign of the encoding/decoding architecture
Solution Approach 2:
The patent implements dynamics by making the weight prediction parameters adaptive rather than static. The system dynamically determines whether to apply weight prediction based on detected brightness changes, and the parameters themselves are derived adaptively from the image data. This dynamic approach allows the system to switch between conventional and weight-based prediction as needed, balancing complexity and accuracy
2Measurement precision
If weight prediction is applied to all blocks regardless of brightness change, then prediction accuracy improves uniformly, but encoding/decoding efficiency deteriorates due to increased processing overhead
Solution Approach 1:
The patent applies local quality by determining weight prediction parameters selectively based on local brightness change characteristics. Instead of uniformly applying weight prediction to all blocks, the system evaluates brightness changes in specific regions and applies weight prediction only where beneficial. This localized approach maintains prediction accuracy in affected areas while avoiding unnecessary processing overhead in areas without brightness changes
Solution Approach 2:
The patent implements partial action by applying weight prediction only to the extent necessary - specifically, only when brightness changes are detected and only to the blocks affected by such changes. The system performs partial derivation of weight prediction parameters based on the actual needs of the image data, avoiding the excessive processing that would result from universal application of weight prediction to all blocks
3Measurement precision
If multiple weight prediction parameter candidates are considered, then prediction accuracy improves through selective optimization, but device complexity increases due to additional parameter management
Solution Approach 1:
The patent applies segmentation by dividing the weight prediction parameter set into distinct candidates (first weight prediction parameter, second weight prediction parameter, and initial weight prediction parameter). Each candidate serves a specific purpose and can be selected based on the particular imaging conditions. This segmented approach allows the system to manage complexity by organizing parameters into discrete, selectable options rather than requiring continuous optimization of all parameters simultaneously
Data Source
AI summary
A method for decoding an image signal according to the present invention may comprise the steps of: obtaining a first prediction sample of a current block by performing inter prediction on the current block; determining one or more weight prediction parameter candidates for the current block; determining a weight prediction parameter of the current block from the one or more weight prediction parameter candidates; and obtaining a second prediction sample of the current block by applying the weight prediction parameter to the first prediction sample. Herein, a number of the one or more weight prediction parameter candidates is variably determined based on a size of the current block.