Encoder Prototype Filter Interpolation for Memory-Limited Larger Transforms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing modulated transform systems require significant memory storage for prototype filters, especially when extending from smaller to larger sizes, which is inefficient and costly, particularly in embedded systems like mobile telephony where memory optimization is crucial.
Innovation Solution
A method to construct a prototype filter of a larger size by inserting calculated coefficients between existing ones, using weighting functions to ensure perfect or almost perfect reconstruction, thereby reducing the number of required coefficients and memory usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the transform size is increased to improve encoding quality and adaptability, then the memory storage requirement for prototype filters increases significantly
Solution Approach 1:
The filter coefficients are segmented into two parts: original coefficients stored in memory and interpolated coefficients generated on-demand using weighting functions. This segmentation allows the system to handle larger transform sizes without proportionally increasing stored coefficient quantity.
Solution Approach 2:
Weighting functions serve as intermediaries that mathematically generate additional filter coefficients from the original set. These weighting functions enable the system to derive coefficients for larger transform sizes without directly storing them, acting as a mediator between the original coefficients and the required extended coefficient set.
2Measurement precision
If more coefficients are stored to support larger transform sizes, then reconstruction accuracy is improved, but memory usage increases
Solution Approach 1:
The original filter coefficients are pre-stored in memory, and the weighting functions are pre-defined. When a larger transform size is needed, the system performs preliminary interpolation calculations using these pre-stored elements to generate the required coefficients, avoiding the need to store all possible coefficients in advance.
Solution Approach 2:
The system changes the parameter of transform size dynamically by interpolating coefficients from the original set using weighting functions. This allows the effective number of coefficients to be adjusted based on the required transform size without changing the actual stored coefficient quantity.
3Productivity
If the encoder is updated to support larger transforms, then the processing capability is improved, but the complexity of the system increases
Solution Approach 1:
The system dynamically adapts to different transform sizes by selectively applying interpolation based on the required size. The complexity is made dynamic rather than static - the system only performs interpolation when larger transforms are needed, and uses original coefficients when smaller transforms suffice.
Solution Approach 2:
Instead of creating entirely new coefficient sets for larger transforms, the system copies and interpolates from the original coefficient set. This copying approach with interpolation maintains processing capability while avoiding the complexity of implementing completely separate coefficient storage for each transform size.
Data Source
AI summary
A method for updating the processing capacity of an encoder or decoder to use a modulated transform having a size greater than a predetermined initial size is provided, particularly, where the encoders or decoders are for storing an initial prototype filter defined by an ordered set of initial size coefficients. A step is provided for constructing a prototype filter of a size greater than the initial size to implement the modulated transform of the greater size by inserting at least one coefficient between two consecutive coefficients of the initial prototype filter.