Adaptive-Weight Generalized Bi-Prediction for Illuminance-Variant Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques, such as bi-prediction, fail to optimally combine motion-compensated prediction signals due to illuminance changes over time, leading to poor performance, especially when illuminance fluctuates rapidly between reference pictures.
Innovation Solution
Implement a generalized bi-prediction framework that uses adaptive weight values to combine multiple motion-compensated prediction signals, allowing for block-level optimization and reduced signaling overhead through a unit-gain constraint and hierarchical weight set construction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing bi-prediction techniques are used to combine motion-compensated prediction signals, then the decoding process is simple, but the prediction accuracy deteriorates when illuminance changes rapidly between reference pictures
Solution Approach 1:
The patent applies parameter changes by introducing adaptive weight values that can be adjusted based on illuminance conditions. Instead of using fixed equal weights in traditional bi-prediction, the system varies the weight parameters dynamically to match changing illuminance levels between reference pictures, thereby improving prediction accuracy without requiring complex real-time analysis
Solution Approach 2:
The patent implements preliminary action by pre-defining a hierarchical weight set at the encoder side that captures various illuminance transition scenarios. These pre-computed weight values are stored and selectively applied during decoding, avoiding the need for complex real-time calculations while maintaining adaptability to different illuminance conditions
2Productivity
If adaptive weight values are used to optimize prediction for different illuminance conditions, then the prediction efficiency improves, but the signaling overhead increases
Solution Approach 1:
The patent uses parameter changes by representing multiple adaptive weight configurations through a compact weight index parameter. Instead of signaling full weight values, the system transmits indices that reference pre-defined weight sets, significantly reducing the bit rate required to convey prediction parameter information while maintaining the ability to adapt to different illuminance conditions
Solution Approach 2:
The patent implements universality by designing a hierarchical weight set that serves multiple illuminance scenarios simultaneously. A single weight index can reference weight configurations suitable for various illuminance transition types, making the signaling mechanism universally applicable across different video content and lighting conditions without requiring separate parameters for each scenario
3Measurement precision
If block-level optimization is implemented for motion-compensated prediction, then the prediction accuracy for local regions improves, but the computational complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the picture into blocks and applying block-level weight selection from the hierarchical weight set. Each block can independently select the most appropriate weight configuration based on its local characteristics, enabling localized optimization without requiring complex global optimization algorithms. The segmentation approach allows parallel processing and reduces computational burden compared to optimizing entire pictures
4Reliability
If multiple motion-compensated prediction signals are combined with adaptive weighting, then the video coding performance improves under varying illuminance, but the processing time increases
Solution Approach 1:
The patent implements preliminary action by pre-computing and storing hierarchical weight sets that capture various illuminance transition patterns. During encoding and decoding, the system only needs to select appropriate pre-computed weights rather than performing complex real-time optimization, significantly reducing processing time while maintaining adaptive performance for different illuminance conditions
Solution Approach 2:
The patent uses parameter changes by representing complex weight configurations through compact indices rather than full parameter sets. This parameter compression allows the system to evaluate and switch between different prediction scenarios rapidly, reducing the time required for weight selection and application while maintaining the ability to adapt to varying illuminance conditions in real-time video processing
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Systems and methods are described for video coding using generalized bi-prediction. In an exemplary embodiment, to code a current block of a video in a bitstream, a first reference block is selected from a first reference picture and a second reference block is selected from a second reference picture. Each reference block is associated with a weight, where the weight may be an arbitrary weight ranging, e.g., between 0 and 1. The current block is predicted using a weighted sum of the reference blocks. The weights may be selected from among a plurality of candidate weights. Candidate weights may be signaled in the bitstream or may be derived implicitly based on a template. Candidate weights may be pruned to avoid out-of-range or substantially duplicate candidate weights. Generalized bi-prediction may additionally be used in frame rate up conversion.