Dependent Quantization State Transition for AV2 Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing dependent quantization methods for video coding, such as those used in versatile video coding (VVC), are not suitable for Alliance for Open Media (AOM) Enhanced Compression Model (AV2) due to significant delay and increased implementation complexity caused by the need to wait for full context-coded syntax elements to be decoded.
Innovation Solution
The proposed solution involves using only the context-coded syntax elements of a previous quantization level to determine the state of a current sample, allowing for the selection of a quantizer without waiting for the entire block to be decoded. This approach eliminates the need for an extra flag indicating the parity of the previous level, reducing the size of the compressed video and improving coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If dependent quantization methods from VVC are applied to AV2, then coding efficiency can be improved, but decoding delay and implementation complexity increase significantly
Solution Approach 1:
The patent segments the syntax elements into two types: context-coded syntax elements and bypass-coded syntax elements. This segmentation allows the decoder to process context-coded elements for determining quantizer state without waiting for all syntax elements to be decoded, thereby reducing decoding delay while maintaining coding efficiency benefits of dependent quantization.
Solution Approach 2:
The patent enables preliminary determination of quantizer state by using only context-coded syntax elements of previous quantization levels, without waiting for the complete decoding of the entire block. This preliminary action allows the decoder to proceed with de-quantization operations earlier, reducing the time loss associated with traditional VVC dependent quantization methods.
2Productivity
If dependent quantization methods from VVC are applied to AV2, then coding efficiency can be improved, but implementation complexity increases
Solution Approach 1:
The patent segments the quantization level representation into context-coded and bypass-coded syntax elements. This segmentation simplifies the implementation by allowing the decoder to use only the context-coded portion for determining quantizer state, avoiding the need to wait for and process the entire syntax element sequence, thus reducing implementation complexity.
Solution Approach 2:
The patent extracts and utilizes only the necessary context-coded syntax elements for determining quantizer state, while ignoring or skipping the bypass-coded syntax elements in this specific processing step. This extraction approach reduces the computational burden and simplifies the decoder architecture compared to processing all syntax elements as required in VVC.
3Device complexity
If an extra flag is added to indicate parity of previous level, then quantizer determination can be simplified, but compressed video size increases
Solution Approach 1:
The patent employs self-service by using the context-coded syntax elements themselves to carry the necessary information for quantizer state determination. The context-coded syntax elements inherently contain the parity information needed for quantizer selection, eliminating the need for separate explicit flags and avoiding additional bits in the compressed video stream.
Data Source
AI summary
A video encoder or decoder reconstructs a block of a video coded according to AOM Enhanced Compression Model (AV2) through dependent quantization. The video encoder or decoder accesses quantized samples associated with the block and processes the quantized samples according to an order for the block to generate respective de-quantized samples. Each quantized sample is associated with context-coded syntax elements and at least one quantized sample is associated with bypass-coded syntax elements. The processing includes obtaining a current quantized sample of the block from the quantized samples and determining a quantizer for the current quantized sample based on a parity of a partial quantization level value represented by context-coded syntax elements of a previous quantized sample. The processing further includes de-quantizing the current quantized sample based on the quantizer to generate a dequantized sample. The video encoder or decoder reconstructs the block based on the de-quantized samples.