Adaptive Quantization for Balanced Pixel-Domain Distortion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding standards, such as H.264, employ uniform quantization in the DCT domain, leading to uneven distortion distribution in the pixel domain, resulting in noticeable artifacts, particularly at the edges of blocks.
Innovation Solution
Adaptive quantization of transform domain coefficients is implemented, where each coefficient is assigned a dynamic quantization step size based on achieving a predetermined average pixel domain distortion, and in-loop deblocking is applied to intra-coded blocks to improve distortion balance and reduce artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If uniform quantization is applied to DCT coefficients, then the quantization process is simple and computationally efficient, but the distortion distribution in the pixel domain becomes highly biased with higher distortion at block edges
Solution Approach 1:
The patent applies different quantization step sizes to different DCT coefficient positions based on their impact on pixel domain distortion. Specifically, coefficients that contribute more to edge pixel distortion (such as high-frequency coefficients) are assigned larger quantization steps, while coefficients affecting center pixels are assigned smaller steps. This local differentiation of quantization quality resolves the contradiction by making the quantization process adaptive to the spatial distribution requirements.
Solution Approach 2:
The patent changes the quantization parameter (step size) from a uniform value to a position-dependent value. The quantization step size is modified as a function of the DCT coefficient position and the desired pixel domain distortion distribution. This parameter transformation allows the system to achieve uniform distortion distribution while maintaining computational efficiency through predefined step size tables.
2Manufacturing precision
If adaptive quantization with position-dependent step sizes is used to balance pixel domain distortion, then distortion distribution becomes more uniform, but the quantization process complexity increases
Solution Approach 1:
The patent pre-calculates and stores optimal quantization step sizes for different DCT coefficient positions in lookup tables before the actual encoding process. These precomputed tables contain the position-dependent step sizes that achieve the desired distortion distribution. During encoding, the system simply retrieves the appropriate step size from the table based on coefficient position, avoiding real-time complex calculations while maintaining uniform distortion distribution.
Solution Approach 2:
The patent introduces precomputed quantization step size tables as an intermediary between the uniform quantization approach and the ideal adaptive quantization. These tables serve as a mediator that translates the complex relationship between DCT coefficients and pixel domain distortion into simple lookup operations, reducing computational complexity while achieving the desired distortion balance.
3Productivity
If quantization is applied to achieve compression, then the bit rate is reduced, but noticeable artifacts appear in the decoded image particularly at block edges
Solution Approach 1:
The patent applies different quantization strengths to different spatial locations by assigning position-dependent quantization step sizes. Regions that are more sensitive to quantization artifacts (block edges) receive smaller effective quantization steps through the mathematical relationship between DCT coefficients and pixel values, while less sensitive regions tolerate larger steps. This preserves compression efficiency while reducing visible artifacts.
Solution Approach 2:
The patent exploits the mathematical relationship between DCT coefficients and pixel domain reconstruction to convert the potential harm of uniform quantization into a benefit. By understanding which coefficients contribute most to edge artifacts, the system deliberately assigns larger steps to those coefficients while compensating with smaller steps to other coefficients, thereby achieving both compression and artifact reduction.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Methods and encoders and decoders for adaptively quantizing transform domain coefficients in image and/or video encoding/decoding. For an image having a number of blocks, each block having N coefficients, the quantization step size for each coefficient is selected dynamically, thus resulting in a quantizer that may apply different step sizes to different coefficients. The selected step sizes for each coefficient are used for those coefficients within all blocks in the image/frame. In one aspect, the step sizes are selected on the basis of balancing pixel domain distortion when the image/frame is decoded.