Video Block SAO Filter Parameter Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional video coding schemes, such as HEVC, face inefficiencies in in-loop filtering due to limited candidate sets of Sample Adaptive Offset (SAO) parameters being encoded, leading to suboptimal compression and reconstruction quality, as most CTUs inherit SAO parameters from a single neighboring CTU, limiting the reduction of coding artifacts like ringing or banding.
Innovation Solution
The method determines a reconstructed version of each block, obtains and refines multiple candidate sets of filter parameters to minimize rate-distortion cost, and encodes these parameters along with an index for each block, allowing for independent optimization of filter parameters for each block, enabling merging or removal of candidate sets based on cost reduction, thereby improving filtering efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If multiple candidate sets of SAO parameters are encoded for each block, then filtering precision and image quality improve, but bitstream size and encoding complexity increase
Solution Approach 1:
The patent merges multiple candidate sets of SAO parameters into a unified structure where common parameters are shared across blocks. Instead of encoding separate parameter sets for each block, the invention combines parameter sets at the CTU level and uses inheritance mechanisms, reducing redundant encoding while maintaining multiple candidate options for filtering optimization.
Solution Approach 2:
The patent applies local quality by allowing different blocks to select from candidate parameter sets based on their specific characteristics. Each block can inherit or override parameters locally depending on its needs, enabling precise filtering adaptation without encoding full parameter sets for every block, thus balancing quality and bitstream efficiency.
2Manufacturing precision
If multiple candidate sets of SAO parameters are encoded for each block, then filtering precision and image quality improve, but device complexity increases
Solution Approach 1:
The patent segments the SAO parameter encoding into hierarchical levels: CTU-level candidate set definition and block-level selection/inheritance. This segmentation separates the complex parameter generation process from the simpler block-level application, reducing overall device complexity while maintaining multiple filtering options.
Solution Approach 2:
The patent performs preliminary action by pre-defining candidate parameter sets at the CTU level before block-level processing. This preliminary preparation reduces the complexity of block-level encoding since blocks only need to select or inherit from pre-computed candidates rather than generating parameters independently, simplifying the encoding device requirements.
3Quantity of substance
If SAO parameters are inherited from neighboring CTUs, then bitstream size reduces, but filtering adaptability and image quality deteriorate
Solution Approach 1:
The patent introduces dynamics by enabling blocks to dynamically select between inheriting SAO parameters from neighboring CTUs or using locally optimized candidate sets. This dynamic selection mechanism allows the system to adapt to local content characteristics while maintaining bandwidth efficiency through inheritance, resolving the contradiction between compact bitstream representation and filtering adaptability.
Data Source
AI summary
An encoding method for encoding blocks of a picture part is disclosed in an embodiment. A reconstructed version of each block of the picture part, called reconstructed block, is determined. A plurality of candidate sets of filter parameters is obtained and an index is determined for each block of the picture part that identifies one of the candidate sets of filter parameters to be used for filtering the block. Each reconstructed block is filtered with the candidate set of filter parameters of the plurality of candidate sets of filter parameters identified by the index. The plurality of candidate sets of filter parameters, the blocks of the picture part and the index for each block are encoded in a bitstream. Advantageously, a plurality of candidate sets of filter parameters are determined such as to minimize a rate-distortion cost for each block of the picture part. For each block of the picture part, an index is determined that identifies one of the plurality of candidate sets of filter parameters to be used for filtering the block such as to minimize a rate-distortion cost for the block. The filter parameters of each candidate set of the plurality of candidate sets are refined using the reconstructed version of the blocks whose index identifies the candidate set.