Coefficient Encoding with Split Bit Planes for High Bit-Depth CABAC
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video data compression and decompression systems face inefficiencies in entropy encoding, particularly with existing techniques like CABAC, which do not fully optimize bit rate and image quality, especially at higher bit depths and low quantization parameters.
Innovation Solution
The implementation of advanced entropy encoding techniques, including CABAC with adaptive context modeling and modified range adaptation, and the use of fixed bits schemes to handle higher bit depths, such as 16-bit systems, by splitting coefficients into most-significant and least-significant parts for efficient encoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If CABAC entropy encoding is used, then encoding efficiency is improved, but bit rate performance deteriorates at higher bit depths
Solution Approach 1:
The patent segments the entropy encoding process into multiple contexts based on coefficient magnitude ranges. Different context models are applied to different magnitude ranges, allowing the encoder to adapt to the statistical characteristics of coefficients at various bit depths. This segmentation enables more efficient probability modeling for higher bit depths while maintaining compression efficiency.
Solution Approach 2:
The patent implements dynamic context adaptation where the entropy encoding parameters are adjusted based on the actual distribution of coefficient magnitudes. The context models are dynamically selected and updated according to the quantization parameter and observed coefficient statistics, enabling optimal encoding performance across varying bit depths and compression requirements.
2Quantity of substance
If quantization is applied to compress video data, then data size is reduced, but image quality deteriorates
Solution Approach 1:
The patent employs adaptive quantization parameter selection that changes based on local image characteristics, frequency bands, and importance of different coefficient positions. By dynamically adjusting the quantization step size and selection, the system achieves better quality-compression tradeoff, preserving important visual information while maximizing compression ratio.
Solution Approach 2:
The patent applies different quantization strategies to different regions and frequency bands of the transform coefficients. Important low-frequency coefficients and coefficients in visually important regions use finer quantization, while less important high-frequency coefficients use coarser quantization. This local differentiation maintains image quality in critical areas while achieving compression overall.
3Quantity of substance
If entropy encoding optimizes for smallest bit representation, then data compression is improved, but encoding complexity increases
Solution Approach 1:
The patent implements a hierarchical entropy encoding approach where not all coefficients are encoded with full entropy coding complexity. Instead, simple run-length encoding or bypass coding is used for coefficients that are likely zero or have simple patterns, while full CABAC is applied only to significant coefficients. This partial application of complex encoding reduces overall complexity while maintaining compression efficiency.
Data Source
AI summary
A data encoding method for encoding an array of data values as data sets and escape codes for values not encoded by the data sets, an escape code including a unary coded portion and a non-unary coded portion, the method including: setting a coding parameter defining a minimum number of bits of a non-unary coded portion; adding an offset value of 1 or more to the coding parameter to define a minimum least significant data portion size; generating one or more data sets indicative of positions, relative to the array of data values, of data values of predetermined magnitude ranges, to encode the value of at least one least significant bit of each data value; generating respective complementary most-significant data portions and least-significant data portions; encoding the data sets; encoding the most significant data portions; and encoding the least-significant portions.