Video Transform Coefficient Coding With Reverse-Scan Level Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards like MPEG-2, H.263, and H.264/AVC face inefficiencies in coding transform coefficients due to fixed code word tables that do not adapt to symbol statistics, leading to suboptimal coding of events with probabilities greater than 0.5 and neglecting inter-symbol redundancies.
Innovation Solution
A method that maps two-dimensional blocks of transform coefficients to one-dimensional vectors using a significance mapping and context-dependent modeling, where significant coefficients are coded with one-bit symbols and magnitudes using unary binarization or exp-golomb coding, allowing for adaptive arithmetic coding that adjusts to actual symbol statistics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If fixed code word tables are used for mapping coding events to code words, then the coding structure is simple and easy to implement, but the coding efficiency is reduced because adaptation to actual symbol statistics is not possible
Solution Approach 1:
The patent implements dynamic code word tables that are adapted during the coding process based on actual symbol statistics. The code word tables are initialized with default values and then updated iteratively as coding proceeds, allowing the system to transition from a static to a dynamic state. This enables the coding system to adapt to the actual distribution of transform coefficient symbols while maintaining a relatively simple implementation structure.
Solution Approach 2:
The patent changes the parameters of the code word tables based on observed symbol statistics. Specifically, the probability estimates for different symbol values are updated during coding, and these updated probabilities are used to adjust the code word assignments. This parameter adaptation allows the system to optimize coding efficiency for the actual content being encoded rather than relying on fixed, pre-determined code words.
2Productivity
If arithmetic coding with non-integer code word lengths is used, then coding efficiency for high-probability events improves, but computational complexity increases significantly
Solution Approach 1:
The patent replaces complex arithmetic coding operations with a simplified binary coding approach. Instead of using full arithmetic coding with non-integer code lengths and complex probability interval calculations, the invention uses binary code words with integer lengths that are determined by adapted probability estimates. This substitution maintains much of the coding efficiency benefit while dramatically reducing computational complexity and making the system more suitable for practical video coding applications.
3Adaptability or versatility
If context-adaptive switching between multiple code word tables is used, then adaptability to different block types improves, but the complexity of managing multiple tables increases
Solution Approach 1:
The patent applies local quality by maintaining different code word table configurations for different types of transform coefficient blocks. Each block type (e.g., intra-coded, inter-coded, different prediction modes) has its own adapted code word table that is optimized for that specific context. This allows the system to adapt locally to the characteristics of each block type while using a unified adaptation mechanism that reduces overall complexity compared to managing completely separate coding systems for each block type.
Data Source
AI summary
The present invention describes a method and an arrangement for coding transform coefficients in picture and/or video coders and decoders and a corresponding computer program and a corresponding computer-readable storage medium, which can particularly be employed as a novel efficient method for binary-arithmetic coding transform coefficients in the field of video coding.For this, it is suggested that, for blocks of (video) pictures containing significant transform coefficients, coding of the transform coefficients takes place in such a way that, for each block in a scan process, the positions of significant transform coefficients in the block and subsequently, in a reverse scan order—starting from the last significant transform coefficient within the block—the values (levels) of the significant transform coefficients are determined and coded.