Transform Coefficient Decoding with Integrated VLC Table Lookup
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video/audio compression standards require significant computational resources for reconstructing transform coefficients, which is inefficient due to the separate steps of variable length decoding and subsequent table look-ups, and are burdened by numerous tables and corner cases.
Innovation Solution
The proposed solution combines transform coefficient decoding with variable length decoding using a two-layer table hierarchy and efficient bit operations, such as peeking multiple run-level pairs and embedding table-switching within the tables, to reduce the number of computation cycles and memory footprint.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If separate steps of variable length decoding and table look-up are used to reconstruct transform coefficients, then decoding accuracy is maintained, but computational complexity and processing time increase significantly
Solution Approach 1:
The patent combines variable length decoding and table look-up operations into a single integrated step by pre-processing the VLC decoding table to include run-level pair information directly in the codeword mappings. This eliminates the need for separate decoding and lookup operations, reducing computational complexity while maintaining decoding accuracy through the unified table structure.
Solution Approach 2:
The patent performs preliminary processing of the VLC decoding table during table construction, pre-calculating and embedding run-level pair mappings for each codeword. This preliminary action eliminates the need for subsequent separate table look-up operations during actual decoding, significantly reducing processing time while preserving accurate coefficient reconstruction.
2Measurement precision
If numerous tables are used to recover transform coefficients in different contexts, then decoding accuracy is maintained, but memory footprint and processing overhead increase
Solution Approach 1:
The patent creates a universal VLC decoding table that handles multiple decoding contexts (intra-coded, inter-coded, different block sizes) within a single table structure. The table is designed to provide context-appropriate run-level mappings through unified codeword entries, eliminating the need for multiple separate tables and reducing memory footprint while maintaining accurate decoding across all contexts.
Solution Approach 2:
The patent merges multiple context-specific tables into a single comprehensive VLC decoding table by consolidating run-level pair mappings for all contexts. This unified table structure reduces the number of tables from dozens to one, decreasing memory usage and simplifying the decoding process while preserving context-aware accurate decoding through the integrated mappings.
3Adaptability or versatility
If conventional two-step decoding process is used, then comprehensive functionality is achieved, but processing speed decreases
Solution Approach 1:
The patent performs preliminary processing to pre-calculate and embed complete run-level pair mappings in the VLC decoding table during table construction. This preliminary action enables single-step decoding that directly outputs transform coefficients without requiring subsequent table look-up operations, significantly increasing processing speed while maintaining comprehensive decoding functionality through the pre-computed mappings.
Solution Approach 2:
The patent merges the variable length decoding step and the table look-up step into a single combined operation by integrating run-level pair information directly into the VLC decoding table entries. This consolidation eliminates the sequential two-step process, doubling the processing speed for coefficient reconstruction while preserving full decoding functionality through the unified table structure.
Data Source
AI summary
Apparatus having corresponding methods and tangible computer-readable media comprise: an input module to receive a bitstream representing data words encoded according to a variable-length code; a peek module to select a peek block comprising a predetermined number of consecutive bits in the bitstream; and a decoder to provide a plurality of the data words based on the peek block.