1. Introduction
Drying is one of the oldest and most important means of preserving
agricultural products. In the past little emphasis was put on quality
attributes of the product. Recently optimisation of the drying process
has become more important due to changed consumer demands on one hand
and increasing energy costs on the other. One of the most important
criteria influencing the consumers' decision is the appearance of a
product, causing a positive or negative sensation.
2. Literature Overview and State of the Art
Visual sensation is a mix of colour, shape and size of a product
(Fernandez et al., 2005). Drying of cellular tissue produces several
chemical (browning and other reactions) and physical (colour, texture,
shape, porosity etc.) changes which are related in some complex ways
(Del Valle et al., 1998).
Lack of analytical understanding of physical changes in the product
during drying typically results in a trial-and-error approach to drying
process development, with associated economic costs (Ho et al., 2002).
Improving drying processes necessitates the knowledge of changes
the product is undergoing during drying in order to find significant
phases. Usually drying time, quality changes which can be easily
determined, e.g. reduction of volume, sometimes product temperature and
energy consumption are determined and used for process optimisation.
Naturally grown materials such as fruits and vegetables are
heterogeneous systems (Brennan, 2006). Aim of a robust drying process
must be the compensation of variations of raw material quality to gain a
more or less homogeneous product (Berget & Naes, 2002; Sahin et al.,
2001; Jorgensen & Naes, 2003).
Most biological products are heat sensitive. Therefore it is
necessary to make sure the product does not exceed certain critical
temperatures during the drying process. For example the ability of wheat
kernels to germinate after drying is preserved up to a drying
temperature of 70[degrees]C. Exceeding this temperature results in the
complete loss of germination ability (Hofacker, 1986). (Lewicki &
Jakubczyk, 2004) showed that up to a certain value (70[degrees]C), there
is no significant influence of drying temperature on mechanical
properties of apples. Exceeding this value a decrease of shrinkage was
observed. They hypothesized that high temperature was responsible for
hardening of the product surface, which in most cases should be avoided.
Colour is an important quality criterion. It does not only
influence the product's subjective appreciation by the consumer but
it c an also be used for the determination of quantitative quality
criteria, which depend on the drying process. This is due to the
correlation between colour, taste and aroma (Morries et al., 1953).
Changes of colour and browning of fruits are effects of a multitude of
chemical reactions like Maillard reactions (Lozano & Ibarz, 1997)
and destruction of pigments. Recent developments allow continuous non
invasive measurement of product colour during the process using low cost
equipment (Sturm & Hofacker, 2008).
During drying shape and size of a particle are changing
continuously. Shrinkage modifies both the dimensions and transport
properties of individual particles as well as the thickness and porosity
of the packed bed in the dryer (Ratti, 1994). Shrinkage does not depend
on water content only. Drying conditions also influence the degree of
shrinkage.
Colour changes are usually measured using a colourimeter. Particles
are taken out of the dryer, colour is measured and afterwards particles
are placed back into the dryer. The drying process is intermitted
particles cool down and have to heat to the initial temperature again.
This means colour changes measured are not identical to those gained in
an uninterrupted process.
Consequently for gaining all information desired, more than one
test has to be carried out. As mentioned above, naturally grown products
are very heterogeneous and therefore no particle or fruit is alike
another. Consequentially all measurements have to be carried out
simultaneously using the same sample to get a valid result. This can
only be achieved using non-invasive optical monitoring systems for
determination of product temperature, colour and shape.
A lot of effort has been made to use machine vision systems when
processing agricultural products. Generally the surface colour or the
dimensions of the product particles are determined and evaluated. it is
assumed that this information allows a quality assessment. Comprehensive
studies of activities in these fields are given by (Brosnan & Sun,
2004) and (Du & Sun, 2004).
New developments in hardware, particularly the availability of
cheap and robust imaging systems allow the development of systems which
derive quality information from images or from a video stream.
(Fito et al., 2004) used image analysis of infrared thermographs
during surface drying of citrus fruits to determine the final drying
time. Drying is finished, when a uniform surface temperature is reached.
it is important that water on the surface is completely removed whereas
the skin of the fruit is not dried. The latter would lead to decay and
citrus quality loss such as peel damage, undesirable flavours
development etc.
Automated colour sorting is commercially used on packing lines for
apples, peaches and several other horticultural commodities (Abbot,
1999).
3. Technological Background of Colour Measurement
Colour of an object can be described by several colour coordinate
systems. Some of the most popular systems are RGB (red, green, blue)
used in colour video monitors, Hunter L,a,b, CIE (Commision
International de l'Eclairage) [L.sup.*][a.sup.*][b.sup.*] and CiE
XYZ. For instrumental measurement CiE and Hunter Systems are of greatest
importance. According to the CiE concepts, the human eye has three
colour receptors and all colours are combinations of those three. Colour
is not a physical value but depends on several physical factors (Figura,
2004). Those factors are source of light, brightness, angle of
illumination, angle of observation and surface condition.
3.1 Human vision
The human eye evaluates visual stimuli according to different
spectral perception functions. However single excitations are combined
to a uniform sensation, colour stimulus specification. Colour vision is
based on red (600 nm), blue (350 nm) and green (550nm) stimulus
perception. Using the measurement of the three human cones'
sensitiveness a physiological colour space can be determined.
Fig. 1 shows the so called tristimulus XYZ curves of the human
colour perception.
[FIGURE 1 OMITTED]
If mixing red, green and blue (RGB) light in different ratios and
intensities most of the visible colour spectrum may be displayed. The
RGB colours are described as additive colours. If all three are
overlaid, light is reflected back to the observer and the colour white
accrues for the observer. In the RGB colour space every RGB component
may have values between 0 (absent) and 255 (maximum).
3.2 [L.sup.*][a.sup.*][b.sup.*]-Colour Space
The [L.sup.*][a.sup.*][b.sup.*] colour space (Fig. 2) is an
international standard, which was developed by the Commission
Internationale d'Eclairage (CIE) in 1976 (Schurr, 2000). It is
device independent and describes the human vision best. All other colour
spaces are contained in it. Colours that are of great importance for
human colour vision are red, green, blue, and yellow.
The [L.sup.*][a.sup.*][b.sup.*] colour space is described in a
three dimensional coordinate system. The L-axis shows the lightness
(0-100), the a axis shows the ratio between green and red (-120...+120)
and the b axis the ratio between blue and yellow ( 120...+120). With
this classification a colour space originates, in which all colours may
be depicted.
This makes the [L.sup.*][a.sup.*][b.sup.*] colour space too
extensive, to be captured by scanners and monitors. RGB colour data
gained can be converted. The transformations necessary are exactly
defined through mathematical operations and described in chapter 5.3.
In contrary to the RGB in the [L.sup.*][a.sup.*][b.sup.*] colour
space a change of lightness does not result in a change of values of
colours. Another advantage of the [L.sup.*][a.sup.*][b.sup.*] colour
space is the consideration of human colour apperception (Wargalla,
2003).
[FIGURE 2 OMITTED]
4. Objectives of this work
For purpose of quality assurance, but also for process control,
quality changes can be determined by continuously measuring the surface
colour of a product. However, some problems have to be resolved:
* The functional link between the information derived from the
image data, the quality criterion to be controlled and the external
process parameters (e.g. temperature, humidity) must be known.
* The system hardware must provide for the images in requested rate
and (constant) quality and must be able to perform the image analysis
and information extraction in an acceptable delay.
Finally, the system should be simple, robust and reliable. The
investment for the system should be as low as possible and comparable to
a conventional system controller.
5. Equipment and Procedures
5.1 Hardware
Experiments were carried out with the experimental setup described
in (Sturm et al., 2009). Air temperature can be adjusted between
5[degrees]C to 160[degrees]C, and dew point temperature between
4[degrees]C to 75[degrees]C. It can be used as an overflow dryer as well
as a single layer dryer. Different drying strategies can be applied: one
stage or multi stage air temperature controlled drying, product
temperature controlled drying or a combination of the strategies
mentioned.
Weight of the sample, pressure drop, surface colour and temperature
are measured continuously. A digital camera (IC-Imaging, Model DFK
31BU03.H) supplies digital images to a PC running under Windows XP. The
visual field of the camera covers one or several product particles
depending of the material's nature. Special efforts were made to
protect the camera from heat and humidity to guarantee for constant
image quality.
Dryer and optical system were protected from daylight to avoid
interference with the artificial illumination of the product samples.
The images are taken continuously and are sent to the computer in pre
set intervals. For the investigations discussed imaging rate was set to
one picture per 5 minutes. Images are stored for later analysis or can
be analysed immediately after exposure for the purpose of automatic
control.
5.2 Procedure
Product temperature was determined continuously using the pyrometer
placed in the drying chamber, inlet air temperature, water bath
temperature, dry bulb and wet bulb as well as drying chamber temperature
were gauged using thermocouples. Air flow rate was detected using a
hot-wire anemometer. All values were transferred to the PLC and from
there to the HMI (Human Machine Interface) WinCC. Weight was measured
continuously using a precision balance and transferred to a LabView
template for data acquisition.
Samples were dried until they reached a dry base water content of
approximately 0.13 g/g.
At the end of drying experiments, samples were weighed and put into
an electric oven for 48 hours at a constant temperature of 70[degrees]C
for the dry matter's determination.
5.3 Image Analysis
Analysis of image data is carried out by a programme specially
developed for this purpose. The area of the image to be analysed is
defined by mouse operation. Information about the background colour and
a white standard is put in. All pixels in the analysis region are
evaluated: if they are part of the product surface, the RGBvalues are
extracted and stored; if they are of the background colour, they are
discarded. Glossy regions are identified by their luminosity. The
minimum-maximum- and mean RGB values are determined. Pixels which
undoubtedly are part of the particle surface are counted to get
information of the particle size and the change of the particle size due
to shrinkage.
The RGB-values are used to calculate the XYZ-Values and
consequentially the [L.sup.*][a.sup.*]b values of the pixels by the
procedures given by (Hunt & Chichester, 1991) and (Leon et al.,
2006):
The first step is to transform the RGB-values in
XYZ-(tristimulus)-values:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (1)
The second step is to transform the XYZ values in
CIE-[L.sup.*][a.sup.*][b.sup.*] values (Schurr, 2000):
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (2)
The values given by CIE for the matrix coefficients and the
constants in the transformation equation for a 2[degrees] observer and
reference RAL white standard are used. Illumination is set on a low
level to avoid reflections.
For determination of quality the Total Colour Difference (TCD) is
calculated from the [L.sup.*][a.sup.*][b.sup.*] values with the function
(Leon et al., 2006)
TCD = [square root of ([DELTA][L.sup.2] + [DELTA][a.sup.2] +
[DELTA][b.sup.2] (3)
6. Results
The following figures show the results of drying experiments with
apples, variety "Jonagold". Figures 3 and 4 depict the drying
process with constant air temperature ([[??].sub.Air] = 85[degrees]C,
60[degrees]C, 35[degrees]C). Developments of product temperatures show
different, significant phases: At the beginning of the drying process,
the temperature rises due to the heat transferred from the air to the
product surface. After this heating phase, the temperature rise is
decelerated, as heat is partly absorbed by the evaporation of the
product humidity. This phase lasts as long as enough moisture can
diffuse to the surface. In the third phase, temperature rise is
accelerated. Almost the whole heat transferred warms the product while
there is only a small amount of water evaporated. This behaviour is
typical for porous materials.
[FIGURE 3 OMITTED]
[FIGURE 4 OMITTED]
As the heat transferred to the product depends mainly on the
temperature of the product surface, this parameter can be used to
control the process: In figures 5 to 7 results for product temperature
control are depicted. Product and drying air temperatures as functions
of time and moisture content, moisture contents as functions of time and
development of colour and shrinkage as functions of moisture content are
shown.
Product temperature was kept constant and air temperature was
adjusted continually. This resulted in a rapid rise of product
temperature to the desired value. Air temperature decrease was analogue
to product temperature rise during air temperature controlled drying.
Variation of maximum air temperature for those tests showed, that a
level of 100[degrees]C should not be exceeded to guarantee a good
product quality. Higher temperature levels can lead to quality losses
due to internal stress in the product.
At the beginning of the process TCD rises quickly to a value of
about 1/3 of the maximum value. Afterwards there is a phase of almost
constant values down to a moisture content of circa X=2. Below this
level TCD changes accelerate again. When air temperature is controlled,
the second phase of rising values starts at a moisture content of X=3.
Changes of quality and temperatures are very different for the
strategies applied. Correlations between temperature changes and quality
changes in the different phases could be seen.
Product temperature control resulted in shorter drying time to
obtain the same moisture content as when controlling air temperature.
Furthermore, product quality--expressed in colour change--was
considerably improved. If comparing identical drying times temperature
level is 13[degrees]C lower when product temperature is controlled. TCD
is significantly lower. The lower temperature level guarantees for a
higher content of heat sensitive components like vitamins and aromas.
[FIGURE 5 OMITTED]
[FIGURE 6 OMITTED]
[FIGURE 7 OMITTED]
Tests carried out showed that two dimensional shrinkage (visible
cross section) cannot be used as quality criterion for products which
shrivel when they shrink. Down to moisture content levels of
approximately X=2 data gained is correct afterwards most of the apparent
shrinkage is due to shrivelling of the particle.
7. Conclusion
Experiments have shown that the continuous measurement of colour
and size of particles as well as development of product temperature
during drying give valuable information on changes, a product undergoes
during the different stages of drying.
The research presented uses colour change vicarious for quality
changes. Chemical and physical properties were not investigated. For a
better understanding of quality changes the functional links between
chemical, physical, optical and sensory properties have to be
determined. The experimental set up does not allow precise prediction of
energy consumption.
Further work is to be done in the field of the determination of the
functional links between colour changes and quality changes. Two more
cameras will be installed in the dryer for investigation of the three
dimensional shrinkage during drying.
DOI: 10.2507/daaam.scibook.2009.50
8. Acknowledgement
The authors acknowledge the German Federal Ministry of Economy and
Technology and the University of Applied Sciences Konstanz for financial
support.
9. References
Abbot, J. A. (1999). Quality measurement of fruits and vegetables.
Postharvest Biology and Technology, Vol. 15, No. 3, (March 1999) pp.
207-225, ISSN 0925-5214
Berget, I & Naes, T. (2002). Optimal sorting of raw materials,
based on the predicted end product quality. Quality Engineering, Vol.
14, No. 3, (March 2002) pp. 459-478, ISSN 0898-2112
Brennan, J.G. (ed.) (2006). Food Processing Handbook, Wiley-VCH,
ISBN 3527307192, Weinheim
Brosnan, T. & Sun, D.W. (2004). Improving quality inspection of
food products by computer vision--a review. Journal of Food Engineering,
Vol. 61, No. 1 (January 2004) pp. 3-16, ISSN 0260-8774
Del Valle, J. M.; Cuadros, T. R. M. & Aguilera, J. M. (1998).
Glass transitions and shrinkage during drying and storage of osmosed
apple pieces. Food Research International, Vol. 31, No. 3, (April 1998)
pp.191-204, ISSN 0963-9969
Du, J. W. & Sun, D. W.(2004). Recent developments in the
applications of image processing techniques for food quality evaluation.
Trends in Food Science & Technology, Vol. 15, No. 5, (May 2004) pp.
230-249, ISSN 0924-2244
Fernandez, L.; Castillero, C. & Aguilera, J. M. (2005). An
application of image analysis to dehydration of apple discs. Journal of
Food Engineering, Vol. 67, No. 1-2, (March 2004) pp. 185-193, ISSN
0260-8774
Figura, L.O. (2004). Lebensmittelphysik (Food Physics), Springer
Verlag, ISBN 3540203370, Berlin Heidelberg
Fito, P. J.; Ortola, M. D. & De los Reyes, R. (2004). Control
of citrus surface during drying by image analysis of infrared
thermography. Journal of Food Engineering, Vol. 61, No. 3, (April 2004)
pp. 287-290, ISSN 0260-8774
Ho, J. C.; Chou, S. K. & Chua, K. J. & Mujumdar, A. S.
& Hawlader, M. N. A. (2002). Analytical study of cyclic temperature
drying: effect on drying kinetics and product quality. Journal of Food
Engineering, Vol. 51, No. 1, (January 2002) pp. 65-75, ISSN 0260-8774
Hofacker, W. (1986). Trocknungsverhalten und Qualitatsveranderungen
von Weizen (Drying Behaviour and Quality Changes of wheat). Dissertation
University Hohenheim
Hunt, R. W. G. & Chichester, E. H. (1991). Measuring colour
(2nd ed.), Wiley, ISBN 0135676789, England
Jorgensen, K. & Naes, T. (2003). A design and analysis strategy
for situations with uncontrolled raw material variation, Journal of
Chemometrics, Vol. 18, No.1, (2003) pp. 45-52, ISSN 0886-9383
Leon, K.; Pedreschi F. & Leon J. (2006). Color measurement in
[L.sup.*][a.sup.*][b.sup.*] units from RGB digital images. Food Research
International, Vol. 39, No 10, (December 2006) pp. 1084-1091, ISSN
0963-9969
Lewicki P.P. & Jakubczyk, E. (2004). Effect of hot air
temperature on mechanical properties of dried apples. Journal of Food
Engineering, Vol. 64, No. 3, (September 2004) pp. 307-314, ISSN
0260-8774
Lozano, J.E. & Ibarz, A. (1997). Colour changes in concentrated
fruit pulp during heating at high temperatures, Journal of Food
Engineering, Vol, 31, No. 3, (February 1997) pp. 365-373, ISSN 0260-8774
Morris, N.J.; Lohmann, I.W. & O'Connor, R.T. &
Freeman, A.F. (1966). Peanut Butter, IV. Determination of color of
peanut butter by spectral reflectance method. Food Technology, Vol. 7,
(1966) pp. 393-396, ISSN 0015-6639
Ratti, C. (1994). Shrinkage during Drying of Foodstuffs. Journal of
Food Engineering, Vol. 23, No. 1, (1994) pp. 91-105, ISSN 0260-8774
Sahni, NS; Isaksson, T. & Naes, T. (2001). The use of
experimental design methodology and multivariate analysis to determine
critical control points in a process. Chemometrics and Intelligent
Laboratory Systems, Vol. 56, No. 1, (January 2001) pp. 105-121, ISSN
0169-7439
Schurr, U. (2000). Handbuch Digitale Bildverarbeitung--Vom Scannen
bis zum Colormanagement--1. Auflage (Digital Imaging--from Scanning to
Colourmanagement, 1. Ed.), dpunkt.Verlag, ISBN 3932588630, Heidelberg
Sturm, B. & Hofacker, W. (2008). Determination of Quality
Changes during the Production of Dried Food Products using Colorimetry
(2008). Annals of DAAAM for 2008 & Proceedings of the 19th
International DAAAM Symposium, Katalinic, B. (Ed.), pp 665, ISBN
978-3-901509-68-1, ISSN 1726-9679, Trnava Published by DAAAM
International, Vienna, Austria 2008
Sturm, B.; Hofacker, W. & Hensel, O. (2009). Automatic Control
of the Drying Process of Biological Materials using Optical Sensors to
acquire Surface Temperature, Color and Shape, Annual International
Meeting of the ASABE, Paper Number 096219 21.06.-24.06.2009, Reno,
Nevada, USA
Wargalla, H. (2003). Farbkorrektur mit Photoshop und Scan
Programmen (Colour Correction with Photoshop and Scaning Programmes),
mitp, ISBN 3826609700
This Publication has to be referred as: Sturm, B[arbara] &
Hofacker, W[erner] (2009). Optical Monitoring and Control of Drying
Processes, Chapter 50 in DAAAM International Scientific Book 2009, pp.
501-512, B. Katalinic (Ed.), Published by DAAAM International, ISBN
978-3-901509-69-8, ISSN 1726-9687, Vienna, Austria
Authors' data: Dipl.-Ing. (FH) Sturm, B[arbara] M.Eng.; Prof.
Dr.-Ing. Hofacker, W[emer], Hochschule Konstanz, Technik, Wirtschaft und
Gestaltung, Brauneggerstasse 55, 78462, Konstanz, Germany,
bstrurm@htwg-konstanz.de, hofacker@htwg-konstanz.de