Material Identification via Spectral Projection Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for material identification using X-ray irradiation and spectrometric analysis are limited in accurately determining the nature of materials independently of their thickness, and there is a need for more precise classification techniques.
Innovation Solution
A method involving irradiation of a material with ionizing electromagnetic radiation, detection of the transmitted radiation using a spectrometric detector, and subsequent projection of the acquired spectrum data using projection vectors determined during a learning phase, allowing for classification of the material into distinct classes based on discriminant criteria.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional spectrometric analysis methods are used to identify materials, then the material nature can be determined, but the identification accuracy is limited and dependent on material thickness
Solution Approach 1:
The patent transforms the spectral data from the original energy domain into a projected space using predetermined projection vectors. This dimensional transformation allows the material classification to be performed in a new space where thickness variations do not affect the classification accuracy, effectively separating material identification from thickness measurement.
Solution Approach 2:
The patent changes the parameter space by projecting the spectral data onto specific projection vectors that are optimized for material classification. By transforming the parameters from the original spectral domain to a projected domain, the method achieves thickness-independent material identification while maintaining high classification accuracy.
2Loss of information
If the full spectral data is used for material classification, then comprehensive information is available, but the complexity of analysis increases
Solution Approach 1:
The patent extracts the essential information for material classification by projecting the full spectral data onto a smaller set of predetermined projection vectors. This extraction process retains the critical material-specific information while eliminating redundant data, thereby reducing analysis complexity without significant loss of classification capability.
Solution Approach 2:
By transforming the spectral data into a lower-dimensional projected space, the patent reduces the computational complexity of the classification process while maintaining the essential material identification information. The projection vectors are designed to preserve the most discriminative features for material classification.
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
This approach enables precise classification of materials regardless of thickness, improving the accuracy of material identification and discrimination between different classes, as demonstrated by the correct grouping of calibration and test spectra in reduced dimensional spaces.
Implementation Method 1
detection of radiation transmitted by said sample by means of a spectrometric detector, said sample being disposed between said irradiation source and said detector
Implementation Method 2
acquisition of a spectrum of the radiation thus detected, called transmission spectrum, representative of the energy distribution of said transmitted radiation
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention is a method for identifying a material contained in a sample. The sample is subjected to irradiation via ionizing electromagnetic radiation, for example X-rays. The sample is inserted between a source emitting said radiation and a spectrometric detector capable of acquiring a spectrum of the radiation transmitted by the sample. The spectrum is subject to different treatment steps so as to enable classification of the material. Said steps are, consecutively: reducing dimensionality, followed by projecting along said predefined projection vectors. Projection makes it possible to establish classification parameters, on the basis of which classification is established.