Dynamic Quantization Rounding Offset for Video Encoding Detail Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding technologies face challenges in achieving transparent visual quality, especially in high bit rate applications, due to the loss of details in smooth regions, as they typically use constant quantization rounding offsets and adjust only the quantization step size, which is not sufficient to preserve fine details like film grain and computer-generated noise.
Innovation Solution
The method involves adjusting the quantization rounding offset dynamically, allowing for flexible selection on a scene, picture, or macroblock level, in conjunction with adjusting the quantization step size, to allocate more bits to smooth regions and preserve fine details, thereby enhancing perceptual quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the quantization step size is adjusted to reduce the number of encoded bits, then the bit rate is controlled, but the visual quality and preservation of fine details deteriorate
Solution Approach 1:
The patent changes the quantization parameters by introducing a dynamically adjustable rounding offset in addition to the quantization step size. This allows independent control of bit rate and quality through two parameters (q and s), enabling fine-tuning of the quantization process to preserve visual quality while controlling bit rate.
Solution Approach 2:
The patent applies different rounding offsets to different regions of the video content based on local characteristics such as texture complexity and gradient magnitude. Smooth regions receive rounding offsets that preserve details, while textured regions use offsets optimized for their specific characteristics, achieving local optimization of visual quality.
2Device complexity
If the quantization rounding offset is kept constant to simplify the encoding process, then the device complexity is reduced, but the ability to preserve fine details in smooth regions deteriorates
Solution Approach 1:
The patent transforms the static rounding offset into a dynamic parameter that adapts to local video content characteristics. The rounding offset is calculated based on regional properties such as gradient magnitude and texture complexity, allowing the encoding process to respond to content variations while maintaining manageable complexity through efficient calculation methods.
Solution Approach 2:
The rounding offset is derived automatically from the video content itself through analysis of local characteristics such as gradient magnitude and texture complexity. The system uses the video data to generate its own quantization parameters without requiring external input or complex manual configuration, achieving adaptive quality optimization autonomously.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
There are provided methods and apparatus for improved quantization rounding offset adjustment for video encoding and decoding. An apparatus includes a video encoder (100) for encoding an input coding unit by determining a residual for the input coding unit corresponding to a difference between the input coding unit and a reference coding unit, applying a transform to the residual to obtain at least one transform coefficient, and quantizing the at least one transform coefficient by selecting a particular quantization step size and a rounding offset.