Intra Prediction Mode Coding Using Residue Differential Signaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding methods, such as the HEVC standard, require maintaining and updating a ranking table for intra prediction modes, which is inefficient and increases the size of the bit-stream due to the need to signal the prediction mode index, especially for blocks far apart in an image with varying characteristics.
Innovation Solution
Transmitting only the prediction mode residue, which is the difference between the current and previous prediction modes, allows the decoder to restore the current mode, eliminating the need for a ranking table and reducing the bit-stream size by assigning shorter codewords to smaller residue values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If a ranking table is maintained to track previously-used prediction modes, then compression efficiency is improved by signaling modes more frequently used at the top, but device complexity increases due to processing cycles and memory requirements for maintaining and updating the ranking table
Solution Approach 1:
The patent extracts only the essential information needed for prediction mode signaling by transmitting the prediction mode index directly without maintaining a separate ranking table. This eliminates the complexity of tracking and updating mode frequencies while still achieving efficient compression by leveraging the statistical property that current modes are likely near the top of what would be a ranking table.
Solution Approach 2:
The system uses the prediction mode indices themselves to infer frequency information without requiring an external ranking table. The decoder can reconstruct the same ranking table independently by counting mode frequencies from the transmitted indices, making the system self-sufficient without additional memory structures.
2Loss of information
If the ranking table is updated for each block based on statistics over the entire previous image, then compression efficiency is improved for frequently-used modes, but loss of time occurs because blocks far apart have different characteristics making the ranking table inefficient until updated with many blocks
Solution Approach 1:
The patent segments the image processing into independent block-level operations where each block's prediction mode is determined locally without waiting for global statistics. This allows parallel processing and eliminates the delay caused by accumulating statistics across the entire image before updating the ranking table.
Solution Approach 2:
The system performs preliminary action by using a default or previously-computed prediction mode for each block without waiting for global statistics to be updated. This allows immediate processing of each block while maintaining good compression efficiency, avoiding the time loss associated with waiting for ranking table updates.
3Measurement precision
If the current prediction mode is signaled explicitly in the bit-stream when not in the MPM set, then measurement precision is maintained by accurately representing the mode, but the size of the bit-stream increases
Solution Approach 1:
The patent changes the parameter representation by transmitting prediction mode indices with variable length codewords where more frequent modes (likely near the top of the ranking table) use shorter codewords. This maintains precise representation of the current mode while reducing the average bit-stream size through differential coding and frequency-based optimization.
Data Source
AI summary
A system and a method for decoding at least a portion of an image includes determining a current prediction mode based on a combination of a prediction mode residue and a function of at least one previous prediction mode and decoding the portion of the image using the current prediction mode.