Method for classifying spectra of objects having complex information content

Patent metadata

Field	Value
Publication number	US-11879778-B2
Application number	US-201716333734-A
Country	US
Kind code	B2
Filing date	Sep 15, 2017
Priority date	Sep 16, 2016
Publication date	Jan 23, 2024
Grant date	Jan 23, 2024

Abstract

The invention relates to a method for classifying spectra of objects having complex information content after recording of the spectra involving the use of a method for preprocessing data and of a method, associated with the data preprocessing, for classification with the calculation of a classifier. After the recording of the spectra and the preprocessing of the spectra, a multiple classification method is thereby performed with at least two different methods for the data preprocessing of the spectra and the method, assigned to the respective data preprocessing, for classification. After the recording and the data preprocessing of the spectra, the following steps are thereby carried out: a calculation of multiple classifiers of the series per type of data preprocessing; a determination of the classifiers of the series with iterative adjustment and validation; a calculation of probabilities of the class association, with all classifiers of the series or classifiers being equally incorporated into the determination of a classification result.

First claim

The invention claimed is:

1. A method for classifying spectra of eggs, comprising:
recording spectral data related to an egg;
preprocessing the spectral data with at least two different spectral preprocessing methods performed in parallel to produce preprocessed outputs;
classifying each of the preprocessed outputs with a respective classifier to produce classification outputs;
calculating a probability related to a class association for each of the classification outputs; and
calculating a final classification result for the egg based on the classification outputs and the probabilities.

2. The method according to claim 1 , wherein a number of classifiers (N G ) is calculated according to the following equation:

N
G

=

v
S

2
⁢

w
S

·

R
S

wherein v S is a number of spectral data points in the spectral data, w S is a doubled half-width of a spectral region in the spectral data, and R S is a number of spectral regions analyzed in the spectral data; and
wherein there is an equal probability of a respective data point in the spectral data being sampled.

3. The method according to claim 2 , wherein data points belonging to the number of the spectral data points (v S ) are weighted.

4. The method according to claim 1 , wherein: at least one of the spectral preprocessing methods is structured such that respectively defined characteristics become prominent and other defined characteristics are suppressed.

5. The method according to claim 4 , further comprising: adding at least one spectral preprocessing method with equivalently defined and equally weighted characteristics to at least one of the spectral preprocessing methods with differently defined characteristics for evaluation.

6. The method according to claim 1 , wherein:
prepossessed spectra are designed as a training set and at least one classifier is defined and validated by the training set.

7. The method according to claim 1 , further comprising:
using at least one method of unsupervised classification or supervised classification to select spectral regions R S or individual wavelength ranges in the spectral data for subsequent analysis.

8. The method according to claim 1 , wherein at least one classifier utilizes a neural network or a linear wavelet transform.

9. The method according to claim 1 , further comprising: using a neural network method or a linear wavelet transform method as a method for classification in classification groups.

10. The method according to claim 1 , wherein: preprocessed spectra are classified using optical molecular spectroscopy.

11. The method according to claim 1 , wherein:
preprocessed spectra are classified using optical molecular spectroscopy selected from absorption, emission, scattering, UV/vis, NIR, IR absorption, fluorescence, and Raman.

12. The method according to claim 1 , wherein at least one of the spectral preprocessing methods comprises any one or more of the following: raw spectra, baseline corrections, normalizations, derivatives, covariance, or Raman spectra.

13. The method according to claim 1 , wherein a calculation of a median or performance of a cluster analysis is carried out for evaluation of the classifiers for a classification result.

14. The method according to claim 1 , further comprising:
acquiring the spectral data via at least one optical device having at least one spectrometer;
storing the spectral data in at least two storage units;
performing the preprocessing of the spectral data by independently preprocessing respective spectral data stored in each storage unit;
separating and configuring the preprocessed spectra data as a training set and as a test set;
calculating the classification outputs using an iterative method and validating the classification outputs;
classifying the preprocessed spectral data of the training set using the classifiers;
placing the spectral data of the training set in a class of object information with an expression of probability for a class association;
calculating a classification result by calculating a median or by performing a cluster analysis to show a probability result for the training set;
classifying the preprocessed spectral data of the test set using the classifiers;
placing the spectral data of the test set in a class of object information with an expression of probability for a class association;
calculating a classification result by calculating a median or by performing a cluster analysis to show a probability result for the test set.

15. The method according to claim 1 , wherein the egg is a bird egg the final classification result is related to a sex of the bird egg.

16. The method according to claim 1 , wherein the egg is a chicken egg and the final classification result is related to a sex of the bird egg.

17. The method of claim 1 , wherein:
each classifier analyzes different spectral data; and
the classification outputs are aggregated and a highest probability associated with the classification outputs is utilized in determining the final classification result.

18. An apparatus for classifying spectra of objects having complex information content, in which the method according to claim 1 is implemented, comprising at least the following:
at least one optical detector having at least one spectrometer or additional detectors for the acquisition and recording of spectral data;
storage units for storing the spectral data;
spectral preprocessors configured to independently preprocess respective spectral data stored in each storage unit;
at least one classificationer for calculating a classifier based on the spectral data, with an incorporation of iterative methods and validating the classifier;
a placer for placing the preprocessed spectral data in at least class of object information with an expression of a probability for a class association;
an evaluator to calculate a classification result, in the form of the median or performing a cluster analysis for determining the probability result.

Assignees

• Univ Dresden Tech

Inventors

• Steiner Gerald
• Preusse Grit
• Koch Edmund
• Galli Roberta
• Schnabel Christian
• Preusse Johanna

Classifications

• G06F18/25

  Fusion techniques · CPC title

• G06F2218/12

  Classification; Matching · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• G01J3/28 Primary

  Investigating the spectrum (using colour filters G01J3/51) · CPC title