SEM Fatigue Striation Analysis with Radon and Spectral Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing computer vision techniques struggle to robustly measure fatigue striation properties in scanning electron microscope images due to large appearance variations and insufficient training data, making it challenging to automate fatigue damage assessment in parts like aircraft components.

Innovation Solution

A computer-implemented method using image processing techniques such as Gaussian filtering, histogram equalization, and Radon transform, combined with spectral analysis, to enhance and analyze fatigue striations in SEM images, determining striation properties like angle and density, and assess fatigue damage by thresholding peak values in power spectra.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If traditional computer vision techniques are used to measure fatigue striation properties, then the measurement process can be automated, but the measurement robustness deteriorates due to large appearance variations and insufficient training data

Engineering Contradiction:
Improveautomation of fatigue damage assessmentVSAvoidrobustness of striation property measurement
Core Design Contradiction:
Extent of automationVSReliability

Solution Approach 1:

The patent replaces traditional machine learning-based computer vision systems with a physics-based signal processing pipeline. Instead of using trained neural networks that require large datasets, the invention uses deterministic image processing operations (autocorrelation, Radon transform, spectral analysis) that are inherently more robust to appearance variations and do not require training data, thereby maintaining automation while improving measurement reliability

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent transforms the image analysis problem from direct feature extraction to spectral domain analysis. By applying autocorrelation to enhance periodicity, then Radon transform to convert spatial information to frequency-space, and finally spectral analysis to identify dominant frequencies, the system extracts striation properties (density, orientation) as spectral parameters that are invariant to appearance variations, resolving the contradiction between automation and robustness

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If manual measurement methods are used for fatigue striation properties, then measurement accuracy can be maintained, but productivity deteriorates due to time-consuming manual analysis

Engineering Contradiction:
Improveaccuracy of striation property measurementVSAvoidspeed of fatigue damage assessment
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent replaces manual visual inspection and measurement with an automated signal processing system that applies mathematical transforms (autocorrelation, Radon, Fourier) to extract striation properties. This substitution maintains measurement precision by using objective mathematical criteria to identify striation characteristics while dramatically increasing productivity through automation, eliminating the time-consuming nature of manual analysis

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent enables the measurement system to automatically determine striation properties without human intervention. The processing pipeline self-adjusts by applying sequential operations (autocorrelation to enhance patterns, Radon transform to analyze orientations, spectral analysis to extract frequencies) and automatically identifies striation density and orientation from the power spectrum, achieving both high precision and high productivity

Inventive Principle:
Principle #25Self-service

3Measurement precision

If complex image processing operations are applied to enhance striation patterns, then measurement precision improves, but device complexity increases

Engineering Contradiction:
Improveaccuracy of striation detectionVSAvoidcomplexity of image processing system
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the image processing task into distinct sequential stages: autocorrelation computation to enhance periodic structures, Radon transform to convert spatial orientations to frequency representations, and spectral analysis to identify dominant frequencies. Each stage performs a specific function that builds upon the previous stage, improving measurement precision through systematic pattern enhancement while keeping individual processing modules relatively simple and well-defined

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3916673B1Method for determining striation properties of fatigue striations and for determining the presence of fatigue damage
Publication Date: 2025.07.16 AIRBUS (SAS)
  • EP3916673B1 patent drawingFigure 1
  • EP3916673B1 patent drawingFigure 2
  • EP3916673B1 patent drawingFigure 3

AI summary

In order to improve the measurement or assessment of fatigue damage to parts a computer-implemented method for measuring striation properties of fatigue striations (26) on a sample surface of a part (12) is proposed. A sample surface (20) is imaged using a scanning electron microscope so as to obtain a sample image (24) potentially containing the fatigue striations (26). A sample image patch (28) potentially containing fatigue striations (26) is selected from the sample image (24) for further processing. After normalizing the sample image patch (28) and enhancing line-like regular structures contained in the sample image patch (28) the resulting normalized image patch (34) is autocorrelated, Radon transformed and spectrally analyzed. The resulting power spectrum (42) of the transformed image patch (38) contains information about the striation properties of the fatigue striations (26) contained in the sample image (24), if any fatigue striations (26) are present. Furthermore, a system for performing the method is proposed.