This document discusses hyperspectral imaging and its applications in the dairy and food industries. Hyperspectral imaging combines imaging and spectroscopy to acquire both spatial and spectral information about a sample. It has advantages over traditional methods as it is non-destructive, fast, and can chemically characterize samples. The document outlines the optical fundamentals, components, analysis, and various applications of hyperspectral imaging in quality control, safety inspection, and process monitoring for foods like cheese, bread, chocolate, and meat. While high cost and data challenges remain, hyperspectral imaging provides possibilities for precision mapping and analysis in the food industry.

1. HYPERSPECTRAL IMAGING IN
DAIRY AND FOOD INDUSTRY
Suchismita Roy
M.Tech(Dairy Technology)

2. Outline
 Introduction
 Hyperspectral Images(HSI)
 Optical Fundamentals of HSI
 Acquisition of HSI
 Components of HSI System
 Analysis of hyperspectral Images
 Application of HSI in food and dairy industry
 Advantages and Disadvantages
 Conclusion

3. Introduction
 Traditionally, assessment of quality and safety involving
human visual inspection, chemical or biological
determination are tedious, time-consuming and
destructive
 These necessitate the need for accurate, fast and non
chemical detection technologies
 Recently, optical sensing technologies have been
investigated as potential tools for non-destructive
evaluation and inspection
(Huang et al.,2014)

4. Contd.
A conventional imaging system is a common technique for
obtaining spatial information of the sample but cannot
identify chemical constituents from a food product
Conventional spectroscopy system is a technique for
evaluating chemical properties or characteristics of food
products but does not provide spatial information
(Sun,2010)

5. Contd.
Hyperspectral is the
technique which can
simultaneously acquire both
spatial and spectral
information of a component
(Sun,2010)

6. Hyperspectral image
 Acquired over the visible
and near-infrared (or
infrared) wavelengths
 Combined and form a
three dimensional
hyperspectral cube, with
two dimensions
describing spatial and
the third one for spectral
information
(Mendoza et
al.,2015)

7. Contd.
 Therefore the measured
spectrum indicates the
ability of the sample in
absorbing or scattering
the exciting light,
representing the inherent
chemical properties of a
sample
(Huang et al.,2014)

8. Optical Fundamentals of HSI
 Food tissues held together by several molecular bonds and
forces
 Water, carbohydrate, fats, proteins are rich in O-H, C-H, N-H
bonds
 Exposed to light waves transmitted, incident energy of
electromagnetic wave changes due to stretching and
vibrations of such bonds
 Leads to detailed fingerprints of food samples
(Huang et al.,2014)

9. Contd.
At macro level transitioning of electromagnetic
wave is expressed as three types of spectroscopic
technique
 Reflected or Transmitted Spectroscopy
 Fluorescence Spectroscopy
 Raman Spectroscopy
(Qin et al.,2017)

10. Reflectance or Transmittance Spectroscopy
 Illumination by Halogen lamp
 Reflected from surface causing specular reflectance
 Multiple scattering by photon leading to diffuse reflectance
 Can work in UV to IR range
(Qin et al.,2008)

11. Fluorescence Spectroscopy
 Excitation of sample by
radiation energy
 UV radiation generally
used in radiation source
 Emitted light observed in
visible spectral range
 Sensitive technique for
measuring subtle
changes in biological
element
(Kim et al.,2012)

12. Raman Spectroscopy
 Passing high energy
monochromatic light
through sample
 Rayleigh Scattering
 Stokes and Anti stokes
Scattering
 Spectrum consisting
peaks indicating
molecular vibration
 Can be used in
quantitative analysis
(McCreery,2010)

13. Acquisition of Hyper Spectral Images
Hypercubes constructed in
2 ways
 Staring imager:
Spatial image field of
view fixed, restricting
light throughput, obtaining
images at one
wavelength after another
Applied in pharmaceutical
quality control
(Lu et al.,2015)

14. Push broom system:
 Acquisition of hypercubes
by relative movement of
object and detector
recording spectrum of
each pixels across a line
of sample
 Application in food
industry well suited to
conveyor belt system
(Lu et al.,2015)

15. Components of HSI System
 Illumination and
acquisition
 Focusing lens
 Wavelength
Modulator
 Detector
(Gowen et al.,2009)

16. Hyperspectral Image Analysis
 Image Calibration
 Spatial and Spectral Preprocessing
 Classification and Regression
 Image Processing
(Gowen et al.,2009)

17. Application of HSI in Food Industry
 Analyze intrinsic and extrinsic component
 Food surface inspection
 Automated poultry carcass inspection
 Moisture distribution in bakery products
 Visualization of sugar distribution in melons
 Analysis of chocolate bar composition
 Bruise detection in apples
 Fat distribution in pork belly and salmon fillet
(Sun,2010)

18. Online inspection of poultry carcass
 Illumination by white LED
 Sensing unit obtains images in 389-753nm
 Analyzing ROI in pairing R580 / R620 spectra giving largest
difference between wholesome and diseased chicken
 99% accuracy in identifying diseased chicken in 140 bpm
line for over 100000 chickens
(Chao et al.,2008)

19. Moisture Distribution in Bread
www.specim.fi accessed on

20. Analysis of Chocolate Bar
www.specim.fi accessed on

21.  Sugar content and
absorption spectra was
investigated
 Absorbance at 676nm
was close to
absorption band of
chlorophyll and
inversely correlated
with sugar content
Sugar Distribution in Melon
(Sugiyama,2010)

22. Application of HSI in Dairy Industry
Application in cheese:
 HSI gives images of the images of the coagulated milk
in cheese production as well as matrix where different
components are embedded
 Can be used to quantify the texture of the different
protein networks of the cheeses
 Can be used in developing sensors
 Can be used in determining transglutaminase activity
(Gowen et al.,2010)

23. Imaging of Cheese Sample
www.newfoodmagazine.com accessed on
20.04.2018

24. Application of HSI in Powder
 Can give image of the constituents and their
distribution in powder
 Gives image of the moisture distribution
 Can be used to determine melamine content in milk
powder
(Munir et al.,2018)

25. Other Application
 Compositional analysis:
Wet chemistry methods of compositional analysis is time and
labor consuming
Mid- and near infrared spectroscopy used for predicting
composition
HSI has added potential of estimating spatial distribution as well
as average composition
(Gowen et al.,2010)
Hyperspectral diffuse reflectance line scanning can be used for
rapid determination of fat by spatially resolved scattering of
light
(Qin et al.,2017)

26. Contd.
In process monitoring
 Homogenization:
Measurement of size distribution of fat globules leading to
control process parameter and point prediction
 Blending:
HSI can spatially map distribution of components allowing
qualitative comparison with control products
(Gowen et al.,2009)

27. Contd.
 Coagulation:
Changes in light scattering
property due modification of
fat globules size and
distribution
HSI is useful for
characterizing the changes
in micelle structure before
gelation of milk and micelle
aggregation during casein
network development
(E’RIC,2010)

28. Control of Techno-functional Properties
 Determines distribution of constituent component
 Turbidity of milk
 Free oil production and meltability of cheese
 Changes that occurred in composition and structure due
to processing
(Gowen et al.,2009)

29. Application in Food Safety
 Early detection of microbial colonies
 Early detection of changes in appearance due to mold
growth
 Extraneous material detection in packaged product
(Gowen et al.,2009)

30. Biofilm Detection
 Fluorescent HSI can be
used to detect biofilm
formation in food contact
surface of pipelines
 The surface containing
biofilm has different spots
from free surface
 HSI could be used to
develop hand held device
for detection even in low
cell population density
(Jun et al.,2010)

31. Advantages and Disadvantages
Advantages:
 No sample preparation
 Non invasive and non destructive method
 Chemical mapping
Disadvantages:
 High cost
 Computational challenges
 Standardized calibration needed
(ElMasry et al.,2010)

32. Conclusion
 Combination of computer vision and optical
spectroscopy
 New possibilities in mapping distribution of constitutes
 Useful in laboratory and online inspection of dairy
products
 Higher precision and faster analysis
 Advances in hardware component, faster analysis and
software will drive the future of hyperspectral imaging