Partition Mode Reordering for Lower-Bandwidth Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding methods suffer from suboptimal bandwidth usage due to fixed ordering of partition modes in video bitstreams, which does not align with the frequency of mode usage, leading to inefficient decoding and encoding processes.
Innovation Solution
Reordering partition modes based on available decoder-side information, using a reference template to generate an optimized list of partition modes, and encoding an index for the best mode to improve decoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a fixed order of mode signaling is used in video bitstreams, then the encoding and decoding process is simple and fast, but the bandwidth usage is suboptimal and decoding efficiency is reduced
Solution Approach 1:
The patent applies dynamics by transitioning from a fixed, static mode signaling order to a dynamic, adaptive ordering that is generated based on reference templates and actual video content characteristics. The mode list is reordered at decoding time according to the specific reference blocks available, allowing the signaling order to adapt to different video scenarios and optimize bandwidth usage for each case.
Solution Approach 2:
The patent changes the parameter of mode signaling order from a fixed constant to a variable that is determined by reference template characteristics. By generating different mode ordering permutations based on the specific reference blocks (e.g., which reference blocks are available, their spatial positions), the system optimizes the bit allocation for mode signaling, reducing the number of bits needed while maintaining decoding efficiency.
2Loss of energy
If partition modes are reordered based on reference templates, then bandwidth requirements are reduced and decoding efficiency is enhanced, but the complexity of the encoding and decoding process increases
Solution Approach 1:
The patent applies preliminary action by pre-defining the set of possible partition modes and their associated reference templates before actual encoding/decoding occurs. The system prepares multiple potential mode orderings in advance based on different reference block configurations, so that during actual processing, the system only needs to select from pre-computed options rather than generating orderings from scratch, reducing real-time computational complexity.
Solution Approach 2:
The patent uses copying by creating reference templates from actual video data and then using these templates to determine mode orderings. The reference templates are copied and reused across different processing steps, allowing the system to leverage existing information (reference blocks) to determine optimal mode signaling without requiring complex real-time analysis of the entire video sequence.
3Ease of manufacture
If a fixed mode signaling order is used, then the system is simple to implement, but it does not align with the frequency of mode usage leading to suboptimal compression
Solution Approach 1:
The patent implements feedback by using the actual video content and reference blocks to dynamically determine the mode signaling order. The system analyzes which reference blocks are available and their characteristics, then feeds this information back into the mode ordering generation process. This feedback mechanism ensures that the mode signaling order aligns with the actual frequency and importance of different partition modes in the given video sequence, optimizing compression efficiency.
Data Source
AI summary
The various implementations described herein include methods and systems for encoding and decoding video. In one aspect, a method of video decoding includes receiving video data that includes a first block from a video bitstream. The method also includes obtaining a set of partition modes and generating a reference template for the first block. The method further includes identifying, from the video bitstream, an index for the reordered set of partition modes, and reconstructing the first block using a first partition mode of the reordered set of partition modes, where the first partition mode corresponds to the index.