Conditionally Nonlinear Transform for Video Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression techniques, such as predictive coding and transform coding, face inefficiencies in processing large-scale signals and non-stationary signals, and fail to utilize human visual and auditory system limitations effectively.
Innovation Solution
A conditionally nonlinear transform (CNT) method is applied to the spatiotemporal volume of a video signal, allowing for independent application to each dimension, combining predictive and transform coding, and utilizing all previously reconstructed signals to obtain optimal transform coefficients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If predictive coding is used to process non-smooth or non-stationary signals, then coding efficiency is improved, but it cannot utilize limitations of human visual and auditory systems
Solution Approach 1:
The patent combines predictive coding and transform coding into a hybrid coding technique that integrates the advantages of both methods. Predictive coding handles non-smooth or non-stationary signals effectively, while transform coding exploits human visual and auditory system characteristics through frequency-domain transformation and quantization, achieving both coding efficiency and perceptual optimization
Solution Approach 2:
The patent creates a composite coding structure that merges predictive coding components (for temporal prediction) with transform coding components (for frequency transformation and quantization). This composite approach allows the system to simultaneously achieve high coding efficiency for complex signals and effective utilization of human sensory system limitations through perceptually optimized quantization
2Measurement precision
If transform coding is used to decompose signals, then important data identification is improved, but it must depend on first available data which reduces prediction signal quality
Solution Approach 1:
The patent applies transform coding to the prediction error signal rather than the original signal. By first performing predictive coding to generate prediction errors, then applying transform coding to these errors, the system achieves accurate data decomposition while maintaining high prediction signal quality, as the transform operates on already-prefiltered residual information
3Productivity
If hybrid coding combining predictive and transform coding is used, then advantages of both techniques are obtained, but device complexity increases
Solution Approach 1:
The patent divides the coding process into distinct functional segments: a prediction stage that handles temporal correlation, a transformation stage that handles spectral decomposition, and a quantization stage that handles perceptual optimization. Each segment is independently optimized and clearly defined, which manages complexity through functional separation while achieving high compression efficiency
Data Source
AI summary
There is provided a method for encoding a video signal based on pixel-correlations on a transform domain, the method comprising: obtaining a first transform coefficient by transforming a pixel value of a target block in a current frame; reconstructing a second transform coefficient for a corresponding block in a previous frame; and obtaining a prediction value of the first transform coefficient based on the reconstructed second transform coefficient and a correlation coefficient.