Transform Coefficient Coding with Significance Mapping and Adaptive Contexts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards, such as 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 coding transform coefficients, then the coding structure is simple and easy to implement, but the coding efficiency is reduced because the tables do not adapt to symbol statistics
Solution Approach 1:
The patent implements dynamic code word tables that are adaptively updated during the coding process based on observed symbol statistics. Instead of using fixed tables, the system dynamically adjusts the code word assignments to match the actual distribution of transform coefficient symbols, thereby improving coding efficiency while maintaining a relatively simple implementation structure.
Solution Approach 2:
The patent changes the parameters of the code word tables by updating their contents based on symbol statistics. The system monitors the frequency of different symbol values and adjusts the code word assignments accordingly, transforming the static parameter set into a dynamic one that adapts to the actual data characteristics.
2Adaptability or versatility
If variable length code words are used for coding events, then the coding flexibility is improved, but events with probabilities larger than 0.5 cannot be coded efficiently
Solution Approach 1:
The patent dynamically adjusts the code word parameters based on the actual probability distribution of symbols. For high probability events, the system assigns shorter code words by updating the code word table during coding, thereby improving efficiency while maintaining the variable length coding flexibility for other events.
Solution Approach 2:
The system incorporates feedback mechanisms that monitor symbol statistics during the coding process and use this information to adjust code word assignments. This feedback loop enables the system to identify high probability events and optimize their coding representation in real-time.
3Device complexity
If a fixed mapping table is used for all transform coefficients in a block, then the implementation is simplified, but inter-symbol redundancies are not utilized
Solution Approach 1:
The patent applies local quality by creating separate code word tables for different regions or contexts within the transform coefficient block. Instead of using a single fixed table for all coefficients, the system tailors the code word assignments to local symbol statistics, thereby utilizing inter-symbol redundancies while maintaining manageable implementation complexity through localized adaptation.
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.