Intra Prediction Mode Restoration for Video Decoding
Solution Overview
Problem
In image compression methods like MPEG and H.264/AVC, intra prediction modes face challenges when adjacent pixel values are unavailable or significantly different, leading to degraded coding efficiency and inability to apply certain prediction modes, resulting in increased distortion and bit requirements.
Innovation Solution
A method that restores intra prediction modes by de-multiplexing bit streams to extract mode information and residual signals, generates reference pixels if necessary, adaptively filters them based on the restored mode, and decodes residual signals to minimize the difference between prediction and original blocks, thereby improving coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the number of intra prediction modes is increased to improve coding efficiency, then the ability to handle diverse image content improves, but the complexity of encoding and decoding increases
Solution Approach 1:
The patent segments the large set of intra prediction modes into multiple groups (e.g., first group and second group). The encoder selects which group to use based on image characteristics, and the decoder reconstructs modes accordingly. This segmentation reduces the immediate complexity burden while maintaining the ability to access diverse prediction modes when needed.
Solution Approach 2:
The patent introduces dynamic adaptation where the encoder and decoder can switch between different intra prediction mode groups based on image content characteristics. The system dynamically adjusts which prediction modes are active and how they are encoded, rather than using a fixed approach. This allows the system to optimize for current image conditions while managing complexity adaptively.
2Measurement precision
If adaptive filtering of reference pixels is applied to improve prediction accuracy, then the similarity between prediction block and original block increases, but the computational complexity increases
Solution Approach 1:
The patent applies adaptive filtering dynamically based on the restored intra prediction mode. The filtering operation is selectively applied only when beneficial, with the filter strength and application determined by the prediction mode itself. This dynamic approach improves prediction accuracy for modes where it matters while avoiding unnecessary computational overhead for modes where filtering provides minimal benefit.
Solution Approach 2:
The patent changes the parameters of the filtering operation based on the intra prediction mode. Different filter coefficients, filter lengths, or filtering strengths are applied depending on which prediction mode is restored. This parameter adaptation allows the system to optimize prediction accuracy for each mode while managing overall computational complexity through selective application.
3Loss of information
If intra prediction modes are restored using adjacent prediction unit modes to reduce bit requirements, then the coding efficiency improves, but the reliability of mode restoration may decrease when adjacent modes are not representative
Solution Approach 1:
The patent uses dynamic mode restoration where the decoder determines which intra prediction modes to restore based on the actual image content and the modes used in adjacent prediction units. Rather than blindly copying adjacent modes, the system adaptively selects which modes to restore and how to reconstruct them, improving reliability while maintaining bit efficiency.
Solution Approach 2:
The patent implements a feedback mechanism where the encoder transmits information about which intra prediction modes are used in adjacent prediction units, and the decoder uses this feedback to intelligently restore modes for the current prediction unit. This feedback allows the decoder to make informed decisions about mode restoration, improving reliability by avoiding blind copying of potentially unrepresentative adjacent modes while still reducing bit requirements compared to transmitting all mode information explicitly.
Data Source
AI summary
Provided is a method that de-multiplexes a bit stream to extract intra prediction mode information and residual signals, restores an intra prediction mode of a current prediction unit using the intra prediction information and intra prediction modes of prediction units adjacent to the current prediction unit, generates reference pixels using one or more available reference pixel if there exist unavailable reference pixels of the prediction unit, adaptively filters the reference pixels based on the restored intra prediction mode, generates a prediction block using the restored intra prediction mode and the reference pixels, decodes the residual signal to generating a residual block, and generates a reconstructed block using the prediction block and the residual block. Accordingly, additional bits resulted from increase of a number of intra prediction mode are effectively reduced. Also, an image compression ratio can be improved by generating a prediction block similar to an original block.