Digital Image Quantization via Multiplication Reciprocals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital image quantization methods require division operations, which are inefficient and often result in inaccurate image compression, leading to artifacts and blurring.
Innovation Solution
Implementing digital image compression using two multiplication operations instead of division, with a method that involves computing scaled reciprocal numbers and compensation values to generate quantization values for each data value in the quantization table.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If division operations are used for quantization, then quantization accuracy is maintained, but processing efficiency deteriorates due to longer clock cycles
Solution Approach 1:
The patent changes the mathematical operation parameter from division to multiplication by introducing scaled reciprocal numbers. The quantization process uses multiplication with pre-computed scaled reciprocal values instead of traditional division operations, achieving both faster processing and maintained accuracy through careful parameter selection and compensation mechanisms
Solution Approach 2:
The patent introduces scaled reciprocal numbers as intermediary values that mediate between the quantization table and the final quantization result. These intermediary multiplication factors enable the replacement of division operations while preserving quantization accuracy through additional compensation steps
2Productivity
If division operations are replaced with single multiplication operation, then processing efficiency improves, but quantization accuracy deteriorates causing image blurring and artifacts
Solution Approach 1:
The patent extracts the reciprocal calculation step from the main quantization loop by pre-computing scaled reciprocal numbers. This separation allows the efficient multiplication operation to be performed without the accuracy losses of single-multiplication approximations, as the complex division is performed once during initialization rather than during image processing
Solution Approach 2:
The patent implements feedback through compensation values that are calculated based on remainder information from the scaled reciprocal number computation. This feedback mechanism adjusts the quantization result to compensate for any inaccuracies introduced by using multiplication instead of division, thereby maintaining quantization precision
Data Source
AI summary
A method, system and computer program product that involves receiving and initializing a digital image. Quantization is preformed on the digital image using at least two multiplication operations. Finally, a compressed version of the digital image is presented for viewing and/or storage or transport.