Coefficient Group Compression Using GCLI Prediction and Quantization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data compression methods are inefficient in reducing the bitrate or data size without compromising quality, especially when the compressed data set size exceeds a data budget, and they do not effectively address the need for simple implementation and reduced meta-data storage.
Innovation Solution
A method that groups coefficients into successive groups, determines the Greatest Coded Line Index (GCLI), performs quantization, and uses entropy encoding to map residues to codes, reducing the data budget required for meta-data and addressing the issue of quantization through horizontal or vertical prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional compression methods are used to reduce data size, then the compressed data set size decreases, but the quality is compromised and meta-data storage requirements increase
Solution Approach 1:
The patent segments the coefficient data into groups of n words, processing each group independently to determine GCLI values. This segmentation allows for efficient compression by focusing on the most significant bits while maintaining quality through selective preservation of coefficient information.
Solution Approach 2:
The patent extracts and codes only the GCLI (Greatest Coded Line Index) values as meta-data, separating this essential information from the full coefficient data. By using unary encoding for GCLIs and entropy coding for residues, the method extracts minimal necessary meta-data while preserving reconstruction quality.
2Quantity of substance
If conventional compression methods are used to reduce data size, then the compressed data set size decreases, but the implementation complexity increases
Solution Approach 1:
The patent employs entropy coding modes (Rice coding, Golomb coding, or unary coding) that are self-adaptive and do not require complex external encoding tables or algorithms. The encoder and decoder use the same simple coding rules, eliminating the need for complex synchronization mechanisms and reducing implementation complexity.
Solution Approach 2:
The patent changes the parameter representation by using GCLI values (indices of highest weight non-zero bits) instead of full coefficient values. This parameter transformation, combined with unary encoding and residue coding, significantly reduces data size while keeping the encoding/decoding logic simple and systematic.
3Measurement precision
If GCLI values are coded using binary encoding, then the coding precision is maintained, but the meta-data volume increases significantly
Solution Approach 1:
Instead of directly encoding GCLI values in binary, the patent inverts the approach by using unary encoding where the code length is proportional to the magnitude of the GCLI. This inversion, combined with predicting GCLIs from previous groups and encoding only residues, dramatically reduces meta-data volume while maintaining the ability to precisely reconstruct original values.
Solution Approach 2:
The patent implements feedback by predicting GCLI values based on previous GCLI values in the sequence. The predictor uses historical data to estimate current GCLI, and only the difference (residue) is encoded. This feedback mechanism reduces the average code length while preserving coding precision for reconstruction.
4Quantity of substance
If quantization is applied to reduce compressed data set size, then the data budget is met, but the quality is compromised
Solution Approach 1:
The patent applies partial quantization by keeping a limited number of bit planes (GCLI bits) and discarding less significant bits. This partial action is sufficient to meet data budgets while maintaining acceptable quality, as the most significant information is preserved through GCLI encoding and residue coding.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
The invention relates to a method for compressing an input data set, wherein the coefficients in the input data set are grouped in groups of coefficients, a number of bit planes,GCLI, needed for representing each group is determined, a quantization is applied,keeping a limited number of bit planes,a prediction mechanism is applied to the GCLIs for obtaining residues, and an entropy encoding of the residues is performed. The entropy-encoded residues, and the bit planes kept allow the decoder to reconstruct the quantized data, at a minimal cost in meta-data.