Machine Learning Video Encoder Quantization Parameter Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern video codecs face increased computational complexity due to improved coding efficiency, leading to longer processing times, despite advancements in coding techniques like rate-distortion optimization and quantization parameter management.
Innovation Solution
A method and apparatus that utilize a machine-learning model to encode image blocks by presenting a derived value from a non-linear quantization parameter, which is used to calculate a Lagrange multiplier for rate-distortion calculations, reducing computational complexity while maintaining coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If improved coding efficiency techniques are used in video encoders, then coding efficiency is improved, but computational complexity increases
Solution Approach 1:
The patent transforms the quantization parameter through a non-linear function to create a derived value that better correlates with rate-distortion characteristics. This parameter transformation enables the machine learning model to make more accurate mode decisions with reduced computational iterations, thereby maintaining coding efficiency while lowering computational complexity
Solution Approach 2:
The patent replaces traditional rate-distortion optimization algorithms with a machine learning model that has been trained on video data. The ML model directly predicts optimal encoding modes based on the derived quantization parameter, substituting complex iterative optimization mechanics with a trained prediction system that achieves similar or better coding efficiency with reduced computational overhead
2Productivity
If more computation time is allocated to achieve improved coding efficiency, then coding efficiency improves, but processing time increases
Solution Approach 1:
The patent performs preliminary training of the machine learning model offline using extensive video data and rate-distortion optimization examples. This preliminary action embeds learned patterns into the model, enabling real-time encoding to use pre-learned knowledge rather than performing complex optimization calculations during actual video processing, thus reducing processing time while maintaining coding efficiency
Solution Approach 2:
The patent creates a simplified computational model that copies the essential decision-making patterns from complex rate-distortion optimization algorithms. The machine learning model replicates the behavior of thorough optimization processes through trained weights and biases, achieving similar coding efficiency outcomes with significantly reduced computational time during actual encoding operations
Data Source
AI summary
Encoding an image block using a quantization parameter includes presenting, to an encoder that includes a machine-learning model, the image block and a value derived from the quantization parameter, where the value is a result of a non-linear function using the quantization parameter as input, where the non-linear function relates to a second function used to calculate, using the quantization parameter, a Lagrange multiplier that is used in a rate-distortion calculation, and where the machine-learning model is trained to output mode decision parameters for encoding the image block; obtaining the mode decision parameters from the encoder; and encoding, in a compressed bitstream, the image block using the mode decision parameters.