TEM Image Crystallinity Analysis via Fourier Transform Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for evaluating crystallinity in TEM images, such as those using image brightness values or deep learning models, face challenges in distinguishing regions of high and low crystallinity due to similar brightness levels and the need for extensive labeled training data.
Innovation Solution
An information processing device that performs a two-dimensional Fourier transform on partial regions of TEM images, clusters frequency strengths, and determines regions of high or low crystallinity without requiring labeled data or thresholding, using vector compression to reduce processing complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If image brightness values are used to evaluate crystallinity, then the evaluation process is simple, but the precision is insufficient because regions with different crystallinity have similar brightness levels
Solution Approach 1:
The patent divides the TEM image into multiple partial regions and performs Fourier transform on each region separately. This segmentation allows the system to capture local structural information that differs between crystalline and non-crystalline regions, enabling precise crystallinity evaluation while maintaining operational simplicity through automated processing
Solution Approach 2:
The patent transforms the image from the spatial domain to the frequency domain using Fourier transform. This dimensionality change converts brightness information into frequency spectrum information, where crystalline regions exhibit distinct peak patterns that can be reliably differentiated from amorphous regions, thereby improving measurement precision without complicating the evaluation process
2Measurement precision
If deep learning models are used to evaluate crystallinity, then the precision can be improved, but the device complexity and data preparation requirements increase significantly
Solution Approach 1:
The patent extracts and analyzes only the essential frequency information from each partial region through Fourier transform, specifically looking for characteristic peak patterns. This extraction approach identifies crystalline regions based on their distinct frequency signatures without requiring complex neural networks or extensive training data, thereby reducing device complexity while maintaining high precision
Solution Approach 2:
The patent replaces the complex mechanical system of deep learning models with a more straightforward mathematical approach using Fourier transform and pattern recognition. This substitution eliminates the need for training data preparation and model optimization, significantly reducing the complexity of implementing the crystallinity evaluation system while achieving comparable or superior precision
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables simple and precise determination of crystallinity at the region level in TEM images, reducing the need for extensive training data and computational resources while effectively differentiating between high and low crystallinity regions.
Implementation Method 1
a conversion section configured to execute a two-dimensional Fourier transform on an image of a partial region for each partial region in the image
Data Source
AI summary
An information processing device acquires an image captured by a transmission electron microscope. The information processing device, for each partial region in the image, executes a two-dimensional Fourier transform on an image of the partial region. The information processing device, based on results obtained by executing the two-dimensional Fourier transform on each of the partial regions, performs clustering of frequency strengths obtained from the results of the two-dimensional Fourier transform. The information processing device determines regions of different crystallinity in the image, based on results of the clustering.