State-Dependent Video Quantization for Accurate Transform Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video coding methods using independent scalar quantization often result in less accurate reconstructed transform coefficients, leading to suboptimal decoded video quality due to insufficient utilization of state information in quantization processes.
Innovation Solution
Implement dependent scalar quantization where the set of reconstruction values for a transform coefficient depends on the values of preceding coefficients, using a weighted sum to adjust transform values and improve accuracy by selecting the closest adjacent state based on current state information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If independent scalar quantization is used, then the quantization process is simple and fast, but the reconstructed transform coefficients are less accurate leading to suboptimal decoded video quality
Solution Approach 1:
The patent applies dynamics by making the quantization process adaptive rather than static. The quantization step size and reconstruction values are dynamically adjusted based on the current state of dequantization, which is determined by the values of preceding transform coefficients. This state-dependent quantization allows the system to adapt to local characteristics of the transform coefficient sequence, improving accuracy without requiring a completely complex new quantization framework.
Solution Approach 2:
The patent changes the parameters of the quantization process based on the current state. Specifically, the set of reconstruction values is changed depending on the state of dequantization, which is derived from preceding coefficient values. This parameter change approach allows the system to select from multiple quantization configurations, improving the accuracy of reconstructed coefficients while maintaining a manageable level of complexity through systematic parameter selection.
2Measurement precision
If multiple quantizers are used with state information, then more accurate shifted values can be obtained, but the quantization process becomes more complex
Solution Approach 1:
The patent segments the quantization process into multiple quantizers, each handling specific states of dequantization. By dividing the overall quantization task into separate quantizer instances that are activated based on the current state, the system can achieve higher accuracy for each segment while the overall complexity is managed through the segmentation structure. Each quantizer is optimized for its specific state, improving overall precision.
Solution Approach 2:
The patent implements feedback by using the current state of dequantization (derived from preceding coefficient values) to determine which quantizer to use and how to adjust the reconstruction values. This feedback mechanism ensures that the quantization process adapts to the actual characteristics of the data being processed, improving accuracy. The state information flows back into the quantization decision, creating a closed-loop system that refines the reconstruction process.
Data Source
AI summary
An example method of video coding includes receiving a video bitstream that includes a plurality of blocks and a plurality of quantized transform coefficients. The method includes deriving a dequantized transform coefficient for a first quantized transform coefficient of the plurality of quantized transform coefficients. The method also includes obtaining an adjusted transform value by applying an offset value to the dequantized transform coefficient, the adjusted transform value based on a quantized transform coefficient level and a current state of dequantization. The method further includes reconstructing a first block using the adjusted transform value.