Spectroscopic characterization of seafood

1. A method comprising:

obtaining, by a device and from a near infrared (NIR) spectrometer, a reflection spectrum of a seafood sample;

performing, by the device, a multivariate pattern recognition analysis of the reflection spectrum;

determining, by the device and based on performing the multivariate pattern recognition analysis, a matching spectrum by comparing the reflection spectrum to a library of known identity spectra corresponding to a plurality of seafood samples,

the library of known identity spectra being generated based on collecting a first plurality of spectra for at least one seafood sample of the plurality of seafood samples and performing an operation on the first plurality of spectra to generate a second plurality of spectra corresponding to the known identity spectra; and

identifying, by the device and based on determining the matching spectrum, the seafood sample.

2. The method of claim 1 , wherein the reflection spectrum is measured at wavelengths between 700 mm and 2500 mm.

3. The method of claim 2 , wherein the reflection spectrum is measured at wavelengths between 950 mm and 1650 mm.

4. The method of claim 1 , wherein obtaining the reflection spectrum includes:

performing repetitive spectral measurements at different locations on the seafood sample; and

averaging the repetitive spectral measurements.

5. The method of claim 1 , wherein obtaining the reflection spectrum includes:

applying an extended multiplicative scatter correction of the reflection spectrum.

6. The method of claim 1 , further including:

computing a Standard Normal Variation of the reflection spectrum,

wherein performing the multivariate pattern recognition analysis of the reflection spectrum includes:

performing the multivariate pattern recognition analysis of the reflection spectrum based on computing the Standard Normal Variation.

7. The method of claim 1 , further including:

performing Savitzky-Golay filtering of the reflection spectrum; and

computing, based on performing the Savitzky-Golay filtering of the reflection spectrum, a first derivative of the reflection spectrum,

wherein performing the multivariate pattern recognition analysis of the reflection spectrum includes:

performing the multivariate pattern recognition analysis of the reflection spectrum based on computing the first derivative of the reflection spectrum.

8. The method of claim 7 , further including:

computing a second derivative of the reflection spectrum,

wherein performing the multivariate pattern recognition analysis of the reflection spectrum includes:

performing the multivariate pattern recognition analysis of the reflection spectrum based on computing the second derivative of the reflection spectrum.

9. The method of claim 1 , wherein obtaining the reflection spectrum includes:

performing a sample-wise normalization of the reflection spectrum.

10. The method of claim 4 , wherein obtaining the reflection spectrum includes:

performing a channel-wise auto-scaling of the reflection spectrum.

11. The method of claim 1 , wherein performing the multivariate pattern recognition analysis includes:

performing a principal component analysis.

12. The method of claim 1 , wherein performing the multivariate pattern recognition analysis includes:

performing soft independent modeling of class analogy.

13. The method of claim 1 , wherein performing the multivariate pattern recognition analysis includes performing at least one of:

a Support Vector Machine analysis,

a Support Vector Classification analysis, or

a Support Vector Regression analysis.

14. The method of claim 1 , wherein performing the multivariate pattern recognition analysis includes performing at least one of:

a Linear Discriminant Analysis (LDA), or

a Quadratic Discriminant Analysis (QDA).

15. The method of claim 1 , wherein performing the multivariate pattern recognition analysis includes performing at least one of:

a TreeBagger Analysis, or

a Random Forest Analysis.

16. The method of claim 1 , wherein performing the multivariate pattern recognition analysis includes performing a Partial Least Square Discriminant Analysis.

17. The method of claim 1 , wherein the seafood sample comprises a skinless fish or fish product.

18. The method of claim 1 , wherein the NIR spectrometer has a weight of less than 100 g.

19. The method of claim 1 , further comprising:

transmitting information associated with the obtained reflection spectrum to a mobile communication device.

20. The method of claim 19 , wherein the information is transmitted via a wireless communication.

21. The method of claim 19 , wherein the information is transmitted via a wired communication.

22. The method of claim 19 , wherein

the information is transmitted from the mobile communication device to a remote server,

the remote server performs particular multivariate pattern recognition analysis, and

the remote server transmits, based on performing the particular multivariate pattern recognition analysis, a result of the particular multivariate pattern recognition analysis to the mobile communication device for display.

23. A method comprising:

obtaining, by a device and from a near infrared (NIR) spectrometer, a reflection spectrum of a seafood sample;

performing, by the device, a multivariate pattern recognition analysis of the reflection spectrum;

determining, by the device and based on performing the multivariate pattern recognition analysis, a matching spectrum by comparing the reflection spectrum to a library of known freshness ratings spectra corresponding to a freshness of the seafood sample,

the library of known freshness ratings spectra being generated based on collecting a first plurality of spectra for at least one seafood sample of a plurality of seafood samples and performing an operation on the first plurality of spectra to generate a second plurality of spectra corresponding to the known freshness ratings spectra; and

providing, by the device and based on determining the matching spectrum, a measure of a freshness of the seafood sample.

24. An apparatus comprising:

a portable NIR spectrometer to obtain a NIR reflection spectrum of a seafood sample; and

an analyzer to:

perform a multivariate pattern recognition analysis of the reflection spectrum;

determine, based on performing the multivariate pattern recognition analysis, a matching spectrum by comparing the reflection spectrum to a library of known identity spectra corresponding to a plurality of seafood samples,

the library of known identity spectra being generated based on collecting a first plurality of spectra for at least one seafood sample of the plurality of seafood samples and performing an operation on the first plurality of spectra to generate a second plurality of spectra corresponding to the known identity spectra; and

identify, based on determining the matching spectrum, the seafood sample.

25. The apparatus of claim 24 , wherein the NIR spectrometer has a wavelength range of between 950 mm and 1650 mm.

26. The apparatus of claim 24 , wherein the NIR spectrometer is absent moving parts.

27. The apparatus of claim 26 , wherein the NIR spectrometer comprises a spectrally laterally variable optical transmission filter coupled to a photodetector array.

28. The apparatus of claim 27 , wherein

the NIR spectrometer further comprises a tapered light pipe, coupled to the spectrally laterally variable optical transmission filter, for collecting light reflected by the seafood sample, and

the NIR spectrometer has a weight of less than 100 g.

29. The apparatus of claim 27 , wherein the analyzer comprises a mobile communication device configured for performing the multivariate pattern recognition analysis of the reflection spectrum.

30. A non-transitory computer-readable storage medium storing instructions for performing the method of claim 1 .