Lossy Data Compression With Tolerance-Based Error Correction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional lossy compression methods often compromise quality at higher compression rates, leading to noticeable distortions and artifacts in images, videos, and audio, which can be unacceptable for technical applications.

Innovation Solution

A method that determines a tolerance range for each input value, generates encoded lossy values, calculates difference values, and adds correction values only when the differences exceed the tolerance range, ensuring the output data set meets specific quality requirements by combining encoded lossy and correction values.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If lossy compression is applied to reduce data size, then compression rate is improved, but quality deteriorates with perceptible distortions and artifacts

Engineering Contradiction:
Improvedata sizeVSAvoidquality
Core Design Contradiction:
Quantity of substanceVSManufacturing precision

Solution Approach 1:

The patent applies different quality levels to different data elements by calculating a distortion threshold for each element and applying correction values only when the difference between original and decoded values exceeds this threshold. This local quality approach ensures that critical data elements maintain high precision while less critical elements can tolerate higher compression, thereby improving overall quality without significantly increasing data size.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent dynamically adjusts the compression parameters by calculating distortion thresholds based on local characteristics of the data. The threshold value is determined by analyzing the data distribution and perceptual sensitivity in different regions, allowing the compression algorithm to adapt its precision requirements locally rather than applying a uniform compression ratio across all data.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If conventional lossy compression is used, then compression rate increases, but distortion and artifacts become noticeable

Engineering Contradiction:
Improvecompression rateVSAvoiddistortion
Core Design Contradiction:
ProductivityVSLoss of information

Solution Approach 1:

The patent implements a feedback mechanism by decoding the compressed data, calculating the difference between original and decoded values, and comparing this difference against a distortion threshold. Based on this feedback, correction values are generated and added back to the decoded data when necessary, thereby reducing distortion while maintaining high compression rates.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

Instead of applying full correction to all compressed data elements, the patent uses partial action by selectively applying correction values only to those elements where the distortion exceeds the calculated threshold. This selective correction approach maintains high compression rates while sufficiently reducing noticeable distortion in critical areas.

Inventive Principle:
Principle #16Partial or excessive action

3Manufacturing precision

If correction values are added to reduce distortion, then quality is improved, but data size increases

Engineering Contradiction:
ImprovequalityVSAvoiddata size
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent applies correction values selectively based on local quality requirements. By calculating distortion thresholds for each data element and applying corrections only where needed, the patent minimizes the amount of additional data required while maintaining quality where it matters most.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent uses partial correction by adding correction values only for data elements that exceed the distortion threshold, rather than applying uniform correction to all elements. This approach improves quality sufficiently without the overhead of correcting every single data point, thereby limiting the increase in data size.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3335423B1Method, device, and computer program product for compressing an input data set
Publication Date: 2025.01.15 UNIVERSITAT STUTTGART
  • EP3335423B1 patent drawingFigure 1
  • EP3335423B1 patent drawingFigure 2
  • EP3335423B1 patent drawingFigure 3

AI summary

The invention relates to a method, to a device (100), and to a computer program product for compressing an input data set. The device (100) comprises: a tolerance range determination unit (20) for determining a tolerance range; an encoder (30) for generating encoded lossy values by lossy encoding of the input values; a decoding unit (40) for generating decoded lossy values associated with the input values, in particular by decoding of the encoded lossy values; a subtraction unit (50) for subtracting each decoded lossy value from the respective associated input value and generating differential values; a comparison unit (70) for comparing each differential value with the respective associated determined tolerance range; a correction value generation unit (60) for generating encoded correction values based on the differential values; and an output data set supply unit (80) for supplying an output data set (90) as a function of a comparison of the differential values with the tolerance ranges and on the basis of the encoded lossy values and/or the encoded correction values.