Title: Food Quality Evaluation Techniques Beyond the Visible Spectrum
1Food Quality Evaluation Techniques Beyond the
Visible Spectrum
- Murat Balaban
- Professor, and Chair of Food Process Engineering
- Chemical and Materials Engineering Department
- University of Auckland
2Definition of Food Quality
- Safety
- - Microbial, chemical
- Nutritional content
- - Micronutrients, macronutrients (composition)
- Physical and Chemical Properties
- - Texture, age, etc
- Appearance and sensory attributes
- - Freshness, ripeness, wholesomeness.
3Context
Measurement of the quality attributes, using
machine vision / image analysis -
Non-destructive - Near real-time -
Reliable - Distribution as opposed to average
values.
4Spectrum
Traditional Machine vision
5Light at different wavelengths interacts with
matter differently
6Advantage of hyperspectral
Spectroscopy
Machine vision
Fast Separates wavelengths Averages the view area
(spatial)
Spatially resolves at pixel level Averages
wavelengths
Hyperspectral Imaging
Separates at pixel level Separates wavelengths.
7(No Transcript)
8Hyperspectral imaging
Wavelengths between 200 and 2500 nm. The food
sample is scanned with many wavelengths.
Can measure moisture, lipids, astaxanthin,
9This gives a 2D view of the sample at each
wavelength.
10Methods
1- Reflectance
Spectrometer or camera
Light source
Sample
11Methods
2- Transmittance
Spectrometer or camera
- Two difficulties
- Thickness affects penetration
- Light disperses
Light source
12Methods
3- Interactance
13Measurement examples
UV
Detection of bones and parasites in fish
(Barnes, 1986)
14Parasites
Manual detection 75 effective
Imaging spectroscopy Depth up to 0.8 cm detected
Speed 1 fillet/sec 40 cm/s
15Composition
Different chemical bonds absorb at different
wavelengths It is possible to scan the food
using many wavelengths, and correlate these with
chemically measured composition. Both the UV and
IR range can be used.
16Composition of cow components
US Patent 4,631,413
17Cocoa powder
Near infrared reflectance factor (R) spectra were
recorded for 60 cocoa powder samples The
spectra were transformed to log (R) versus l, and
to the second derivative of log (1/R) versus
wavelength for correlation with compositional
data Linear stepwise regression techniques were
used to determine the optimum l and other
parameters for predicting chemical
constituents The ratio of second derivatives of
log (1/R) measured at two characteristic
wavelengths.
18Composition of cocoa powder
Kaffka et al., 1982
19Fish
ElMasry and Wold, 2008
20Hyperspectral water and fat analysis
Atlantic halibut
Catfish
Cod
Herring
Mackerel
Saithe
21NIR cold smoked salmon
22Oyster Composition
Oysters were homogenized Composition was
measured by wet chemistry, then scanned high
throughput 250300 samples can be analyzed for
moisture, fat, protein and glycogen each day.
Brown 2011
23Protein
Moisture
Glycogen
Fat
24Meat Ageing
(Firtha, 2012)
25(Firtha, 2012)
26Methods of Data Analysis
- Chemometrics
- These methods include (not exclusively)
- partial least squares (PLS) regression,
- multiple linear regression (MLR), and
- principal component analysis (PCA).
Pork quality
27Summary
In addition to visible light analysis (size,
color, shape, texture, etc) UV and IR regions can
also be used for quality evaluation. These
include composition, specific objects (e.g.
parasites, or bones), tenderness. Advantages
Use of multiple wavelengths allow more insight
into the materials Disadvantages Multiple
wavelengths require complex chemometric analysis.
28Thank you
JenOptik 60 mm macro Lens UV-VIS-IR
Nikon D300S UV and IR filters removed