Video Encoder Rate Control via Macroblock Class Bit Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video encoding methods using adaptive quantization struggle to evenly distribute quantization noise across macroblocks, leading to visible artifacts in decoded video, especially in smooth areas where moderate quantization spread is insufficient.
Innovation Solution
The method involves dividing macroblocks into classes based on statistical activity and measuring bit usage to determine optimal quantization levels, allowing for intelligent adjustment of quantization parameters, particularly lowering them for smooth macroblocks to reduce noise visibility.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If high quantization values are used to reduce bit usage, then bit rate is reduced, but quantization noise becomes more visible
Solution Approach 1:
The patent applies different quantization values to different macroblock types locally. Smooth macroblocks receive lower quantization values to minimize noise visibility, while complex macroblocks receive higher quantization values to save bits. This local differentiation resolves the contradiction by optimizing the noise-bitrate tradeoff for each region independently.
Solution Approach 2:
The patent dynamically changes the quantization parameter based on macroblock characteristics. By measuring variance in luminance samples and classifying macroblocks accordingly, the system adjusts quantization values adaptively, allowing higher quantization for complex areas and lower quantization for smooth areas, thus resolving the noise-bitrate contradiction.
2Quantity of substance
If quantization values are increased to reduce bit usage in smooth macroblocks, then bit rate decreases, but quantization noise becomes more noticeable
Solution Approach 1:
The patent specifically targets smooth macroblocks for differentiated treatment by measuring luminance variance and classifying them separately. These smooth regions receive lower quantization values to prevent noise visibility, while other regions can use higher quantization values, thus resolving the contradiction locally where it matters most.
Solution Approach 2:
The patent performs preliminary classification of macroblocks based on luminance variance before applying quantization. By identifying smooth macroblocks in advance through variance measurement, the system can pre-assign appropriate lower quantization values to these regions, preventing noise visibility issues before encoding occurs.
3Stability of the object's composition
If the spread in linear quantization is limited to 4:1 ratio, then encoding stability is maintained, but quality in smooth macroblocks remains insufficient
Solution Approach 1:
The patent removes the uniform 4:1 spread limitation by applying local quality differentiation. Smooth macroblocks are identified through variance measurement and granted lower quantization values that would otherwise violate the global spread limit. This local exception resolves the contradiction by prioritizing quality in smooth areas where noise is most visible.
Solution Approach 2:
The patent dynamically adjusts quantization parameters based on measured luminance variance, allowing the spread ratio to exceed 4:1 for smooth macroblocks. By changing the quantization parameter adaptively rather than enforcing a fixed ratio, the system resolves the contradiction between stability and quality in smooth regions.
Data Source
AI summary
A method for video encoding is disclosed. The method generally includes the steps of (A) dividing a plurality of first macroblocks into at least two classes based on a plurality of first statistics, (B) measuring a respective number of bits used to encode the first macroblocks within each of the classes and (C) based on the measuring in step B, determining a quantization level in at least one of a plurality of second macroblocks that have yet to be encoded.