Digital Image Encoding Method Using Localized Coefficient Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital image compression methods predict all signs globally, leading to degraded prediction quality and inefficient compression due to interdependence between sign predictions.
Innovation Solution
A method where description elements are ordered based on associated scores and predicted individually using the best combination of values from previously processed elements, allowing for improved prediction quality and efficient compression by selecting a subset of elements to be predicted and encoding their differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all signs are predicted globally using conventional methods, then compression is attempted, but prediction quality degrades due to interdependence between sign predictions
Solution Approach 1:
The patent segments the prediction process by dividing signs into two independent groups: selected signs that are predicted individually based on local characteristics, and non-selected signs that are encoded explicitly. This segmentation eliminates the interdependence problem by treating each sign independently rather than predicting all signs globally, thereby improving prediction quality while reducing the complexity of the prediction process.
Solution Approach 2:
The patent applies local quality by selecting signs for prediction based on their individual characteristics (amplitude threshold and position in scanning order) rather than applying a uniform prediction method to all signs. Each selected sign is predicted independently using its local properties, which improves prediction accuracy while avoiding the complexity of global prediction methods.
2Productivity
If a predetermined number of coefficients are selected for prediction based on amplitude and block size, then compression ratio increases, but prediction quality is degraded due to global prediction influence
Solution Approach 1:
The patent segments the set of coefficients into two distinct groups: selected coefficients that undergo prediction and non-selected coefficients that are encoded explicitly. This segmentation allows the system to achieve high compression ratio by predicting only the most significant coefficients while maintaining prediction accuracy by avoiding the influence of other coefficients on the prediction process.
Solution Approach 2:
The patent applies partial action by selecting only a subset of coefficients for prediction rather than predicting all coefficients. This partial prediction approach achieves sufficient compression ratio improvement while maintaining prediction accuracy, as the selected coefficients are those most likely to benefit from prediction based on their amplitude and position characteristics.
3Measurement precision
If signs are predicted by testing all possible combinations to maximize likelihood criterion, then prediction accuracy improves, but computational complexity increases significantly
Solution Approach 1:
The patent segments the prediction process into two stages: first selecting coefficients based on simple amplitude and position criteria, then predicting only those selected coefficients individually. This segmentation avoids the need to test all possible combinations globally, significantly reducing computational complexity while maintaining prediction accuracy through the use of locally optimized prediction for each selected coefficient.
Solution Approach 2:
The patent applies partial action by predicting only a subset of coefficients rather than all coefficients. This reduces the search space from all possible combinations to only the combinations of selected coefficients, dramatically reducing computational complexity while maintaining sufficient prediction accuracy for the most significant coefficients.
Data Source
AI summary
A method for encoding a digital image divided into a plurality of blocks of pixels. The method includes, for a current block: processing the current block to provide a set of description elements; selecting a subset of description elements to be predicted; ordering the description elements of the subset into an ordered sequence; and encoding the elements of the sequence. Encoding includes scrolling the elements of the sequence and includes, for a current element: selecting a combination of predicted values of the description elements of the current sequence from a plurality of possible combinations based on a predetermined cost criterion and from the second element, values of description elements previously processed of the initial sequence; predicting the current element of the sequence by its value in the selected combination; and encoding an indicator representative of a difference between the actual value of the current element and its predicted value.