Adaptive Reference Filtering for Bi-Predictive Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding methods, such as those in the MPEG-4 AVC standard, are inadequate in handling non-translational mismatches in bi-predictive frames across different reference lists, leading to inefficiencies in coding efficiency due to the inability to separately consider and compensate for different types of mismatches from List 0 and List 1.
Innovation Solution
The implementation of adaptive reference filtering (ARF) for bi-predictive pictures in multi-view coded video, where reference filter coefficients are determined based on luma and chroma values of pixels grouped by depth, disparity, and motion, allowing for the estimation and application of multiple sets of filters to compensate for mismatches between reference frames from List 0 and List 1.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single set of interpolation filters is applied to both List 0 and List 1 reference pictures, then the device complexity is reduced, but the coding efficiency deteriorates due to inability to separately compensate for different types of mismatches from different reference lists
Solution Approach 1:
The patent divides the reference picture filtering process into separate handling for List 0 and List 1. Different interpolation filters are selected and applied independently to reference pictures from each list based on their specific mismatch characteristics, rather than using a single unified filtering approach for both lists.
Solution Approach 2:
The patent applies different filtering characteristics to different reference lists based on their local requirements. List 0 and List 1 reference pictures receive tailored interpolation filters suited to their specific mismatch types (e.g., temporal mismatches for List 0, inter-view mismatches for List 1), optimizing coding efficiency for each list's particular characteristics.
2Productivity
If adaptive interpolation filtering is applied to all reference pictures, then the coding efficiency is improved, but the computational complexity and processing time increase
Solution Approach 1:
The patent applies adaptive interpolation filtering selectively rather than universally. Filters are applied only to specific reference pictures that benefit from them, with the selection based on mismatch detection between the current picture and reference pictures. This partial application reduces processing overhead while maintaining coding efficiency benefits where needed.
Solution Approach 2:
The patent dynamically changes filtering parameters (filter type, filter strength) based on the detected mismatch characteristics between reference pictures and the current picture. When mismatches are detected, appropriate adaptive filters are applied; when mismatches are minimal, filtering may be reduced or skipped, optimizing the balance between coding efficiency and processing time.
Data Source
AI summary
There are provided methods and apparatus for adaptive reference filtering of bi-predictive pictures in multi-view coded video. An apparatus includes an encoder (100) for encoding a current picture as a bi-predictive picture. The encoder (100) performs adaptive filtering of at least one reference picture to respectively obtain at least one filtered reference picture, and bi-predictively codes the current picture using the at least one filtered reference picture. The at least one reference picture is a picture wherein at least one sample thereof is used for inter-prediction either of subsequent to the least one sample being applied to an in-loop filter or in an absence of the at least one sample being applied to an in-loop filter. Reference filter coefficients for the adaptive filtering are determined responsive to at least one of luma values and chroma values of pixels grouped based on at least one of depth and disparity and motion.