Guided Transcoding Using Delta Quantized Coefficients
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Guided transcoding approaches for video sequences face challenges in balancing storage requirements and computational complexity, particularly in efficiently encoding and decoding quantized coefficients, which are crucial for adaptive bit rate streaming to meet varying bandwidth and device capabilities.
Innovation Solution
A method for encoding and decoding delta quantized coefficients based on estimated quantized coefficients derived from intra mode and inter motion information, reducing the need for storing actual transform coefficients and enhancing encoding efficiency by exploiting correlations between delta and estimated coefficients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If actual transform coefficients are stored for all LQ versions of video sequence, then video quality is maintained, but storage requirements increase significantly
Solution Approach 1:
The transform coefficients are segmented into two parts: actual quantized coefficients (stored minimally) and delta quantized coefficients (stored instead). This segmentation allows the system to store only the differences rather than complete coefficient sets, reducing storage requirements while maintaining the ability to reconstruct full-quality video when needed.
Solution Approach 2:
The invention extracts and stores only the essential side information (motion vectors, intra mode information, coding unit structure) and delta quantized coefficients, removing the need to store complete transform coefficient sets for all LQ versions. This extraction approach maintains video quality capability while dramatically reducing storage requirements.
2Reliability
If complete LQ versions are encoded and stored, then video quality is maintained, but computational complexity increases
Solution Approach 1:
The system performs preliminary encoding to extract side information (motion vectors, intra mode information, coding unit structure) and store it for future use. When LQ versions are needed, this pre-extracted side information is reused, avoiding the need to perform complete encoding computations again, thus reducing computational complexity while maintaining video quality.
Solution Approach 2:
Instead of storing complete LQ video versions, the invention stores a copied/derived set of side information that can be reused to reconstruct LQ versions on-demand. This copying approach maintains the ability to deliver quality video while significantly reducing the computational burden of encoding multiple complete versions.
3Quantity of substance
If side information is stored without actual picture data, then storage space is reduced, but encoding speed decreases
Solution Approach 1:
The invention changes the parameters being stored from complete picture data to compressed side information and delta quantized coefficients. This parameter transformation reduces storage space requirements while the efficient encoding of delta coefficients (using context-based arithmetic coding) minimizes the impact on encoding speed, achieving a balance between storage reduction and encoding efficiency.
4Ease of manufacture
If delta quantized coefficients are encoded without using estimated coefficients, then encoding simplicity is maintained, but bit rate increases
Solution Approach 1:
The encoding process uses feedback from estimated quantized coefficients (derived from side information) to guide the encoding of delta quantized coefficients. The encoder compares actual coefficients with estimated coefficients and encodes only the differences, using the estimation accuracy to optimize the bit rate allocation. This feedback mechanism reduces bit rate while maintaining encoding simplicity through context-based arithmetic coding.
Data Source
AI summary
Delta quantized coefficients of a pixel block in a picture in a first representation of a video sequence are encoded based on information derived from estimated quantized coefficients for the pixel block. The delta quantized coefficients represent a difference between actual quantized coefficients of the pixel block derived by encoding at least a portion of the picture in the first representation of the video sequence and the estimated quantized coefficients. The estimated quantized coefficients represent a difference between a reconstructed block of a corresponding picture in a second representation of the video sequence and a prediction block obtained based on intra mode information and/or inter motion information derived by encoding the picture in the first representation of the video sequence.