Method and device for spectral reconstruction

• US 4,642,778 A
• Filed: 03/02/1984
• Issued: 02/10/1987
• Est. Priority Date: 03/02/1984
• Status: Expired due to Fees

• Current Assignee: ADVANCED RESEARCH & TECHNOLOGY INSTITUTE
• Original Assignee: Indiana University Foundation
• Inventors: Honigs, David H., Hieftje, Gary M.
• Primary Examiner(s): Krass, Errol A.
• Assistant Examiner(s): Dixon, Joseph L.
• Application Number:
  US06/585,575
• Time in Patent Office: 1,075 Days
• Field of Search: 364/498, 364/497, 364/484, 364/485, 364/576, 364/572, 364/716, 364/726, 364/724, 343/55 A, 73/659, 324/310, 324/312, 382/17, 356/300, 356/303, 356/402, 356/425, 250/226, 340/722
• US Class Current: 702/23
• CPC Class Codes: G01J 2003/284 Spectral construction
• CPC Class Codes: G01J 2003/2866 Markers; Calibrating of scan
• CPC Class Codes: G01J 3/28 Investigating the spectrum using colour filters G01J3/51
• CPC Class Codes: G01N 21/3577 for analysing liquids, e.g. polluted water

Abstract

A method is presented for spectral reconstruction which comprises first obtaining component concentrations of a series of mixtures, which component concentrations constitute a set of reference values. The spectral value of each member of that series of mixtures is measured at a first wavelength, which spectral value possesses contribution from individual components in the mixture, whose individual contribution is unknown. The spectral values for the series of mixtures are then mathematically cross-correlated with the component concentrations in said series of mixtures, at the first wavelength, thereby obtaining the spectral contribution for said components at that first wavelength. The spectral value of each member of that series of mixtures is measured at the second wavelength, which spectral values are then cross-correlated with component concentrations in the series of mixtures thereby obtaining the spectral contribution for said components at the second wavelength. This operation is repeated for a series of wavelengths until the spectrum of the desired component in the mixture is reconstructed. Devices are presented which implement the methods for spectral reconstruction of the present invention.

Claims

1. A method for spectral reconstruction comprising the steps of:

(a) obtaining the concentrations of a first component in a series of mixtures of components, the concentrations of the first component constituting a set of reference values;

(b) measuring spectral values for each mixture in the series of mixtures at a series of wavelengths;

(c) cross-correlating, at a first wavelength, the spectral values for each mixture in the series of mixtures with the corresponding reference values for the one component at the first wavelength to obtain the spectral contribution for the one component at the first wavelength;

(d) outputting the spectral contribution for the one component at the first wavelength;

(e) repeating steps c and d at wavelengths other than the first wavelength, until the spectrum of the one component is obtained.

2. A method as claimed in claim 1 wherein the cross-correlating step comprises calculating:

##EQU5## wherein n is the number of mixtures in the series of mixtures, x is the mixture index, a(x) is the spectral value of the mixture x at the first wavelength, b (x) is the reference value of the first component in the mixture x, a is the average spectral value of all mixtures in the series of mixtures at a first wavelength, b is the average fractional concentration of the first component in the series of mixtures, and C'"'"'_ab is the value of the cross-correlation function between a(x) and b(x).

3. A method as claimed in claim 2 wherein the spectrum is a near-infrared reflectance spectrum.

4. A method as claimed in claim 1 wherein the cross-correlating step comprises calculating:

##EQU6## wherein n is the number of mixtures in the series of mixtures, x is the mixture index, a(x) is the spectral value of the mixture x at the first wavelength, b(x) is the reference value of the first component in the mixture x, a is the average spectral value of all mixtures in the series of mixtures at a first wavelength, b is the average fractional concentration of the first component in the series of mixtures, and C"_ab is the value of the cross-correlation function between a(x) and b(x).

5. A method as claimed in claim 4 wherein the spectrum is a near-infrared relectance spectrum.

6. A method as claimed in claim 1 wherein the spectrum is a near-infrared reflectance spectrum.

19. The method of claim 1 or 13 further comprising repeating the reconstruction of spectra for additional components of the mixtures.

7. A device for spectral reconstruction:

(a) means for obtaining the concentrations of a first component in a series of mixtures of components, the concentrations of the first component constituting a set of reference values;

(b) spectrometer means for obtaining spectral values for each mixture in the series of mixtures;

(c) data storage means for storing the reference values and spectral values;

(d) data input means for supplying the reference values and spectral values to the data storage means;

(e) spectral reconstructor means for cross-correlating the corresponding reference values and spectral values stored in the data storage means, thereby providing a spectral contribution for the one component;

(f) data output means for outputting the spectral contribution for the one component.

8. A device as claimed in claim 7 wherein the means for spectral reconstruction cross-correlates the reference values and spectral values according to the equation:

##EQU7## wherein n is the number of mixtures in the series of mixtures, x is the mixture index, a(x) is the spectral value of the mixture x at the first wavelength, b(x) is the reference value of the first component in the mixture x, a is the average spectral value of all mixtures in the series of mixtures at a first wavelength, b is the average fractional concentration of the first component in the series of mixtures, and C'"'"'_ab is the value of the cross-correlation function between a(x) and b(x).

9. A device as claimed in claim 8 wherein the spectrometer means is a near-infrared reflectance spectrometer.

10. A device as claimed in claim 7 wherein the means for spectral reconstruction cross-correlates the reference values and spectral values according to the equation:

##EQU8## wherein n is the number of mixtures in the series of mixtures, x is the mixture index, a(x) is the spectral value of the mixture x at the first wavelength, b(x) is the reference value of the first component in the mixture x, a is the average spectral value of all mixtures in the series of mixtures at a first wavelength, b is the average fractional concentration of the first component in the series of mixtures, and C"_ab is the value of the cross-correlation function between a(x) and b(x).

11. A device as claimed in claim 10 wherein the spectrometer means is a near-infrared reflectance spectrometer.

12. A device as claimed in claim 7 wherein the spectrometer means is a near-infrared reflectance spectrometer.

13. A method for reconstructing the spectrum of a specific component from a series of multicomponent mixtures, each mixture capable of exhibiting a composite spectrum, the method comprising:

(a) obtaining the concentration of the specific component in each mixture in the series;

(b) measuring a composite spectral value for each mixture in the series at a plurality of wavelengths; and

(c) for each of the plurality of wavelengths, cross-correlating an array comprised of the measured composite spectral value for each mixture in the series of mixtures with a correspondingly ordered array of the concentration of the specific component in each mixture in the series of mixtures to generate a reconstructed spectral value associated with the specific component.

14. The method of claim 13 wherein the step of measuring the composite spectral value for each mixture in the series comprises:

(a) selecting a plurality of wavelengths at which it is desired to have a reconstructed spectral value for the specific component; and

(b) generating an array of composite spectral values at each of thc selected wavelengths, each array of composite spectral values comprising the measured spectral value of each mixture at one of the selected wavelengths.

15. The method of claim 14 wherein the step of cross-correlating the array of composite spectral values for each mixture with the correspondingly ordered array of concentrations of the first component comprises calculating:

##EQU9## wherein, n is the number of mixtures in the series, x is the series index, d is the displacement from the current x index, a(x) is the measured spectral value of the xth mixture at a specific one of the selected wavelengths, b(x) is the concentration of the first component in the xth mixture, and C_ab (d) is the value of the cross-correlation function a(x) and b(x) at the displacement d.

16. The method of claim 14 wherein the step of cross-correlating the array of composite spectral values for each mixture with the correspondingly ordered array of concentrations of the first component further comprises compensating the value of the cross-correlative function for experimental noise and interference.

17. The method of claim 16 wherin the step of compensating the value of the cross-correlation function comprises cross-correlating the array of composite spectral values for each mixture with the correspondingly ordered array of concentrations of the first component according to the calculation:

##EQU10## wherein, n is the number of mixtures in the series, series index, a(x) is the measured spectral value of the xth mixture at a specific one of the selected wavelengths, a is the average spectral value of all samples at the same specific one of the selected wavelengths, b(x) is the concentration of the first component in the xth mixture, b is the average fractional concentration of a given component, the denominator in the equation is the variance of b(x), and C'"'"'_ab is the compensated and normalized value of the cross-correlation function a(x) and b(x).

18. The method of claim 17 wherein the step of cross-correlating the array of composite spectral values for each mixture with the correspondingly ordered array of concentrations of the first component further comprises compensating for mixtures in which the fractional concentration of another component in the mixture influences the concentration of the specific component for which the spectral reconstruction is desired.