Surface Analyzer Clustering Accuracy via Histogram Threshold Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional clustering methods in elemental mapping analysis using EPMA can lead to false cluster detection due to uneven or specific distributions of data points on scatter diagrams, affecting clustering accuracy.
Innovation Solution
A surface analyzer that adjusts the distance threshold in density-based clustering using distribution information from signal values on binary or ternary scatter diagrams, and excludes data points with small summed signal values in ternary diagrams to prevent false cluster detection, employing hierarchical density-based spatial clustering with noise and data point selection based on histogram analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If density-based clustering is applied to scatter diagram data points, then automatic cluster extraction is achieved, but false clusters are detected due to uneven or specific data point distributions
Solution Approach 1:
The patent applies preliminary action by performing data point exclusion before clustering. Data points with small summed signal values in ternary scatter diagrams are excluded in advance using histogram analysis, preventing them from forming false clusters during the subsequent density-based clustering process
Solution Approach 2:
The patent changes parameters by adjusting the distance threshold in density-based clustering based on the distribution characteristics of signal values. The parameter adjustment unit modifies the distance threshold parameter to adapt to the specific data distribution, improving clustering accuracy while maintaining automation
2Measurement precision
If distance threshold is adjusted using distribution information, then clustering accuracy is improved, but device complexity increases due to additional processing units
Solution Approach 1:
The patent applies universality by making the parameter adjustment unit utilize the same distribution information that is already generated for other processing purposes. The histogram analysis and distribution information are reused across multiple functions including data point exclusion and distance threshold adjustment, avoiding redundant processing
Solution Approach 2:
The system applies self-service by automatically adjusting the distance threshold based on the inherent distribution characteristics of the data. The parameter adjustment unit autonomously determines appropriate parameters from the data's own distribution information without requiring external intervention or complex manual configuration
Data Source
AI summary
An object of the present invention is to improve the accuracy of clustering by avoiding detection of false clusters when automatically clustering points on a scatter diagram. A surface analyzer according to a first aspect of the present invention includes a measurement unit (1-2, 4-8) configured to acquire a signal reflecting a quantity of a plurality of components or elements that are analysis targets at a plurality of positions on a sample (3), a scatter diagram generation unit (92) configured to generate a binary scatter diagram based on a measurement result by the measurement unit, a clustering unit (94) configured to perform clustering of points in the binary scatter diagram using a method of a density-based clustering, and a parameter adjustment unit (93) configured to adjust a distance threshold by utilizing distribution information on a signal value of the components or the elements on either axis in the binary scatter diagram, the distance threshold being one of parameters to be set in the density-based clustering.