Trends in Food Science & Technology 39 (2014) 76e84
Viewpoint
Computer vision
technology for
real-time food quality
assurance during
drying process
Mortaza Aghbashloa,*,
Soleiman Hosseinpoura and
Mahdi Ghasemi-Varnamkhastib
a
Department of Agricultural Machinery Engineering,
Faculty of Agricultural Engineering and Technology,
University of Tehran, Karaj, Iran (Tel.: D98 261
2801011; fax: D98 261 2808138; e-mail:
maghbashlo@ut.ac.ir)
b most difficult or even impossible. Therefore, it is of interest
Department of Mechanical Engineering of Biosystems,
Shahrekord University, Shahrekord, Iran
Drying process causes many changes in the mechanical,
sensorial, and nutritional properties of food products. One
main challenge in the production of dried food products
with acceptable shape, size, color, and texture is to monitor
and control their appearance in real-time manner. Currently,
there is an increasing demand for real-time approaches such
as computer vision technology to monitor and control the
food quality indicators including shape, size, color, and
texture during drying process. This note briefly describes the
potential application of computer vision system in the moni-
toring and controlling of the food drying process in order to
enhance the dried product quality and identifies prospects
for future investigations.
Background
Drying process is an important and extensively used oper-
ation in food industry in order to prevent (or inhibit) the
growth and activity of micro-organisms, to make it
* Corresponding author.
0924-2244/$ - see front matter Ó 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tifs.2014.06.003
available throughout the year and at all locations, to extend
the expected shelf life of product, and to reduce the weight
and bulk of food for reducing the charges of transportation
and storage (Aghbashlo, Mobli, Rafiee, & Madadlou,
2013). However, there will be a significant loss in the nutri-
tional, functional, mechanical, and sensorial parameters of
food products if the drying process is performed inappro-
priately. Additionally, the acceptability and marketability
of dried food is greatly put through its uniformity in accor-
dance with the structure, shape, size, color, texture, and fla-
vor. Since the visual appearance and sensible aroma are the
most important criteria in preferring a new product by
consumers.
Traditionally, the quality of dried food are usually eval-
uated off-line by a number of quality parameters such as
moisture content, structural shrinkage, surface color, rehy-
dration ratio, chemical compositions, viscoelastic and me-
chanical properties, and internal microstructure on end-
product samples. However, discrimination of failure cause
and abatement of the process failure using these traditional
off-line evaluations and subsequent empirical models are
to apply advanced, in situ, real-time inspection tools,
capable of in-process monitoring of the organoleptic, me-
chanical, and nutritional attributes using the indirect visual
properties measurements during drying process. In other
words, these characteristics must be precisely monitored
and controlled not only for the products quality but also
for a better understanding of the ill-defined drying process
and complex underlying mechanisms. Furthermore, food
drying process is a quite complex and dynamic phenome-
non, whose undergoing mechanisms are not yet completely
understood. It could be attributed to the simultaneous heat
and mass transfer phenomena, case-hardening, surface
cracking, and severe physiochemical and biological
changes occurred during drying process. As well, drying
process is susceptible to the environmental and random
modifications as results of unsteady ambient conditions.
These can lead to a great product quality loss with a large
deviation in the organoleptic attributes of finished products.
Hence, the conventional measurement techniques cannot be
accurately applied for real-time monitoring and controlling
of food drying process. This research field becomes even
more important due to a trend today in food industry
from batch processes towards continuous processes, in
which the request for in-process quality feedback control
 M. Aghbashlo et al. / Trends in Food Science & Technology 39 (2014) 76e84 77

are increasing for producing the healthy dried foods.

Furthermore, non-homogeneity, anisotropy, and non-
uniformity of the biological materials hinder successful
application of the previously developed empirical models
for precise real-time monitoring and automation. Therefore,
it is essential to measure the food quality parameters in an
in-process successive manner to achieve the efficient drying
process in food industry.
On the other hand, according to Quality by Design strat-
egy (QbD), quality must be retained, protected, and even
built-not examined-into the foodstuffs through processing
units. A QbD development process may include several
steps as clearly shown in Fig. 1.
Obviously, real-time measurement techniques have an
important role in the QbD approach. These techniques
are commonly non-destructive and fast and are divided
into three categories i.e. in-line, on-line, and at-line, ac-
cording to Food and Drug Administration (FDA, 2004).
The differences between off-line, at-line, on-line, and
in-line measurement approaches are manifested in
Fig. 2. At-line process measurement is often performed
by sampling from the process and analyzing in close vi-
cinity to the process apparatus within the timescale of pro-
cessing. In the on-line approach, the gathered sample from
process stream is diverted to the measurement equipment
via an automatic sampling device and is frequently re-
turned to the process stream. In-line process measurement
(invasive or non-invasive) is carried out without with-
drawing sample from the process stream, by direct placing
of the analyzers into the process stream (Burggraeve,
Monteyne, Vervaet, Remon, & Beer, 2012). It can be
concluded that the in-line and on-line approaches are
less time-consuming and have a big potential for devel-
oping a feedback controller based on the monitoring of
the organoleptic attributes.
Nowadays, real-time monitoring of food appearance by
computer vision system has become a main issue in food
industry, since it is consistent, efficient, and cost effective
alternative over the off-line destructive methods. This
approach is a rapid, safe, reliable, and non-destructive tech-
nique, which requires no sample withdrawing and can be
Fig. 2. An illustrative schematic for indicating the difference between
off-line, at-line, on-line, and in-line measurement techniques.
applied as in-line, at-line, on-line and off-line measurement
tools for the shape, size, color, and texture analyses in food
processing operations. One key attribute that enables image
processing suitable for qualitative measurements during
food processing is the relationship between the product
quality attributes and the appearance of foods including
physical structure, color, and visual texture. Thus, a rapid
image-based screening technique providing information
regarding the organoleptic properties would be useful in
food processing operations.
Fig. 3 manifests the analogy between the biological and
artificial vision systems. Obviously, the computer vision
system tries to mimic the processing and behavior of bio-
logical systems as much as possible.
In general, the main steps in computer vision technology
can be divided into five steps as follows:
(1) Image acquisition to provide a digital image by one or
several image sensors;
(2) Pre-processing including deblurring, image compres-
sion, re-sampling, noise reduction, contrast enhance-
ment, and scale space representation to improve
captured image;

Fig. 1. QbD steps for food quality assurance.

78 M. Aghbashlo et al. / Trends in Food Science & Technology 39 (2014) 76e84
Fig. 3. Analogy between the biological and computer vision systems.
(3) Detection/segmentation to segregate the digital image
to separate non-overlapping zones,
(4) Feature extraction operations to provide the characteris-
tics of objects; and
(5) High-level processing to classify a detected object into
various categories or to estimate specific parameters
(Dowlati, Guardia, & Mohtasebi, 2012).
It is worth mentioning that there is no universal agree-
ment on terminology of image analysis due to various
mathematical algorithms to manipulate the image data for
different scopes (Dowlati et al., 2012). However, for non-
invasive and non-intrusive assessment of foodstuffs during
various operations via visual assessment of their quality
three main stages including capturing, processing, and
analyzing of the images must be taken into account. Never-
theless, the feasibility of applying in-line measurement ren-
ders the computer vision system a competitive solution for
studying the organoleptic attributes of foodstuffs during
various unit operations. Fortunately, the in-line monitoring
of the foodstuffs appearance provides useful insight on the
process fulfillment and the instantaneous products quality,
which mark an urgent aid for plant managers and key infor-
mation for model-based control devices. Therefore, for both
scientific and industrial food processing operations, in-line
monitoring of the organoleptic and sensorial attributes has
established as a versatile approach for process understand-
ing, monitoring, controlling, and automating.
Shape and size monitoring
Shrinkage occurs often during drying of food, on which
the viscoelastic matrix shrinks back into the voids previ-
ously occupied by the water (Hosseinpour, Rafiee, &
Mohtasebi, 2011). Consequently, this will become visible
as changes in the shape and structure i.e. density, porosity,
specific volume of the dried products (Bonazzi &
Dumoulin, 2011). These changes are also affected by the
moisture transport mechanisms and the partial pressure of
water (Koç, Eren, & Kaymak Ertekin, 2008). It is worth
noting that the intensive shrinkage in most cases negatively
impresses the consumers’ preference and acceptance
(Mayor & Sereno, 2004). On the other hand, the rehydra-
tion capability and surface cracking are determinant factors
in the optimization of drying condition to produce the high
quality dried products, which is drastically influenced by
shrinkage (Aghbashlo, Kianmehr, & Hassan-Beygi, 2010).
Hence, monitoring and controlling the structural shrinkage
during food drying via real-time measurements techniques
are of major importance.
Several authors made attempts to develop a universal
concept for the prediction of fruit and vegetable volume
shrinkage by taking into account that the total mass of
moist material contains dry solids, water, and air
(Krokida & Maroulis, 1997; Rahman, Perera, Chen,
Driscoll, & Potluri, 1996; Zogzas, Maroulis, & Marinos-
Kouris, 1994). Accordingly, some empirical correlations
developed to estimate the bulk density of fruits and vegeta-
bles as function of moisture content (Ghiaus, Margaris, &
Papanikas, 1997; Koç et al., 2008; Lozano, Rotstein, &
Urbicain, 1983; Madamba, Driscoll, & Buckle, 1994).
Michailidis, Krokida, Bisharat, Marinos-Kouris, and Rah-
man (2010) comprehensively discussed about the conven-
tional methodologies and various empirical models for
measurement and estimation of density, shrinkage, and
porosity of foodstuffs. Apparent shrinkage can be measured
by simple off-line methods including geometric dimension,
liquid displacement, gas pycnometer, and solid displace-
ment for most of foodstuffs during thermal and non-
thermal processing (Michailidis et al., 2010). Measuring
of shrinkage by these techniques is time-consuming and
sometimes causes deterioration of end-product, if adjusting
the operating condition is postponed. In addition, shape and
size changes can be reflected in many parameters such as
average volume, average area, perimeter, Feret diameter,
roundness, elongation, shape of edges, and equivalent
diameter of products (Hosseinpour et al., 2011). This
 M. Aghbashlo et al. / Trends in Food Science & Technology 39 (2014) 76e84
means that it is impossible to obtain more information
about the shape and size of products during drying process
by the use of traditionally destructive measurements and
even with off-line measurements of shrinkage, porosity,
and density. Furthermore, sampling and preparation not
only can lead to significant analytical errors but also can
disturb mechanisms of an ongoing process. These in turn
restricts the widespread application of off-line approaches
for real-time monitoring and precise in-process closed-
loop control based on the shape and size monitoring.
Recently, computer vision has been extensively em-
ployed to analyze the effect of drying condition on the
shape and size changes in both off-line (Fernandez,
Castillero, & Aguilera, 2005; Mayor, Moreira, & Sereno,
2011; Mulet, Garcia-Reverter, Bon, & Berna, 2000; Ramos,
Silva, Sereno, & Aguilera, 2004; Yan, Sousa-Gallagher, &
Oliveira, 2008) and in-line (Campos-Mendiola et al., 2007;
Hosseinpour et al., 2011; Yadollahinia & Jahangiri, 2009)
modes. These studies have demonstrated that the computer
vision technology is a useful tool for both off-line and in-
line tracking of the shape and size during drying process
with an acceptable accuracy. However, due to sample with-
drawing and its preparation, the information obtained using
the off-line computer vision is not always comparable with
the in-line measurements. In-line computer vision does not
require sample removing and it can be applied to track the
shape and size variations during food processing. Gener-
ally, real-time computer vision has indicated to be a prom-
ising tool for monitoring the shape and size changes during
drying process, not only capturing the volume shrinkage,
porosity, and bulk density but also obtaining the additional
useful information on the shape and size variations.
Color monitoring
Color is one of the key characteristics of a dried product,
since it can profoundly reflect the chemical and biochem-
ical reactions occurring during food processing. It is
notable that the rates of these reactions strongly depend
on the drying techniques and the processing conditions
(Kerr, 2007). Therefore, tracking of color during foodstuffs
processing may be used as an indirect way for estimating
the chemical and biochemical reactions in the easier and
cheaper procedure than the commonly utilized off-line
chemical analysis (Maskan, Kaya, & Maskan, 2002). On
the other hand, a slight deviation from initial color gives
a feeling of freshness to the products, while inappropriate
color negatively influences the processed foodstuffs
marketability. As well, some off-colors can be established
as results of incorrect drying conditions due to the loss or
degradation of pigments such as chlorophylls and caroten-
oids (Kerr, 2007). Generally, the success of foodstuffs dry-
ing as a routinely used unit operation in food industry is
attributed to the fact that the foodstuffs retain their original
color, shape, size, and texture to a great extent.
Traditionally, the visual and spectrometric techniques
are often used for color measurements (Figura &
79
Teixeira, 2007). The visual color measurement was
commonly performed by perceiving a sample without
measuring tools at controlled lightening conditions and
comparing the results with a set of color standards. But
this simple approach is found to be inappropriate for
most industrial applications due to its great operator-
dependency and often poor repeatability. The spectrometric
techniques like colorimeter measure the absorption of
known wavelengths by the sample at controlled specified
lightening situation. However, these techniques give quite
unrepresentative insights and additionally have limited
application for the global analysis, since the measuring sur-
face must be uniform and rather small (Mendoza &
Aguilera, 2004). Furthermore, these methods are slower
in food processing media because of the sampling and prep-
aration and may fail in suggesting an appropriate process-
ing condition for real-time process manipulations.
Although, the off-line procedures are widespread in food
industry for color evaluation, non-destructive, fast, and reli-
able measurements have been established for successive
monitoring of the food color variations during various
unit operations using the low cost computer vision technol-
ogy. Nowadays, thanks to the enormous development in the
computer power and the image processing algorithm, com-
puter vision has currently come within reach for foodstuffs
color assessment. Literature survey evidences that the
remarkable attempts has been made to investigate the color
changes of foodstuffs during drying process using the com-
puter vision technology in off-line (Fernandez et al., 2005;
Jinorose, Prachayawarakorn, & Soponronnarit, 2010;
Mohebbi, Akbarzadeh-T, Shahidi, Moussavi, & Ghoddusi,
2009) and in real-time (Chen & Martynenko, 2013;
Hosseinpour, Rafiee, Mohtasebi, & Aghbashlo, 2013;
Nadian, Rafiee, Aghbashlo, Hosseinpour, & Mohtasebi,
2014) modes. Results of the previous surveys have indi-
cated that the computer vision technology frequently pro-
vides precise information on the color kinetics of
foodstuffs during drying process in both off-line and real-
time manners. However, the off-line destructive approaches
and even off-line computer vision systems are considered
improper for actual process condition due to the inability
to provide the real-time information and the highly
operator-dependency and poor repeatability. Nevertheless,
the in-line computer vision is a rapid and non-destructive
approach that is well suited for color analysis of food dur-
ing different operations. Thus, the in-line computer vision
system has a big potential to enter drying technology not
only for color measurements but also for better understand-
ing and controlling of the progression of the chemical and
biochemical reactions.
Texture monitoring
Food texture can be defined as those attributes of a food
that are perceptible by touch in the mouth and with the
hands. Food texture is another important parameter, which
is significantly influenced by drying process and drying
80 M. Aghbashlo et al. / Trends in Food Science & Technology 39 (2014) 76e84
factors. Generally, dried foodstuffs are more rigid, and
become tighter at lower water contents (Kerr, 2007).
Texture is considered as one of the main quality attributes
of the dried products as it influences their organoleptic
and sensorial attributes. Therefore, monitoring and control-
ling of the foods texture is an important task not only to
improve dried product quality and shelf life but also to
enhance the consumer preference. According to
Phadungath (2003), texture is a multidimensional class of
attributes of the food materials. The textural characteristics
of foodstuffs and popular terms with emphasis on the me-
chanical characteristics are illustrated in Table 1.
In the case of food processing, three-point bending test,
compression and puncture test, universal testing machines,
texture profile analysis (Texturometer), and extrusion
testing procedure are usually used to assess the foodstuffs
texture in the off-line mode. The conventional destructive
texture analysis methods can provide precise results on
the strength and consistency of processed foods. However,
they cannot provide in-process information that may be
critical for in-line monitoring and controlling of the
above-mentioned attributes. A perfect procedure for
measuring the texture would be real-time, non-destructive,
capable of operating at different drying condition and even
in industrial scales, and be able to cover a wide range of
food variety.
Recently, several non-destructive approaches such as ul-
trasound technology, near-infrared spectroscopy, and
hyper-spectral imaging have been successfully applied to
measure the textural attributes of processed foods (Chen
& Opara, 2013). In the ultrasonic technique, very short ul-
trasonic pulse-waves are transmitted into the object of in-
terest to characterize the materials. Changes in elastic
attributes of materials inside a processing unit are the basic
measurement concept behind the ultrasonic monitoring.
The desired information regarding the textural attributes
of material can be extracted by distinguishing the reflected
and attenuated signals using a chemometric technique
(Whitaker et al., 2000). Near-infrared spectroscopy utilizes
infrared region of the electromagnetic radiation with longer
wavelengths (800e2500 nm) than those of the visible light
and a frequency range of about 300 GHz up to 430 THz.
The moisture content, chemical state, and textural
Table 1. Relationships between mechanical characteristics and
popular nomenclatures (Phadungath, 2003).
Primary Secondary Popular terms
parameters parameters
Hardness Soft, firm, hard
Cohesiveness Brittleness Crumbly, crunchy, brittle
Chewiness Tender, chewy, tough
Gumminess Short, mealy, pasty, gummy
Viscosity Thin, viscous
Elasticity Plastic, elastic
Adhesiveness Sticky, tacky, gooey
properties of the material within processing equipments
can be effectively monitored using the infrared spectros-
copy due to the sensitivity of absorbance in this region to
variations in the moisture content, chemical state, and
textural properties (da Silva, Butzge, Nitz, & Taranto,
2014). It is worth mentioning that the absorbance in the
near-infrared zone originates from molecular overtone
and combination vibrations. Furthermore, the near-infrared
imaging system (hyper-spectral imaging spectrometer) has
been recently introduced with capability of providing
simultaneously spectral and spatial insights of the materials
to be processed. It should be noted that the hyper-spectral
imaging has a major potential to find applications in drying
technology for monitoring the textural properties. These
tools can be used to improve process understanding and
to minimize processing loss in the real-time non-intrusive
procedure without sample preparation. However, these
techniques suffer from several drawbacks such relative
high equipment cost, requiring sophisticated software for
data analysis, and formulation-specific nature of their mea-
surement. It seems that other innovative techniques
including electrical capacitance tomography, Raman spec-
troscopy, microwave resonance technology, X-ray densi-
tometry, and computer vision technology that are
repeatedly applied in pharmaceutical and chemical indus-
tries, have a great potential for real-time texture monitoring
during foodstuffs processing. Briefly, the electrical capaci-
tance tomography is a technique to image the varying
permittivity distribution in the interior of an insulating me-
dium from external capacitance measurements outside the
boundary of the medium, using the sensing electrodes usu-
ally install around its outer-boundaries (Watzenig & Fox,
2009). The captured images are thus reconstructed into to-
mograms with different approaches to obtain insights on
the electrical characteristics of the process volume. It seems
that the textural properties of foodstuffs significantly influ-
ence their dielectric properties and, accordingly, permit-
tivity distribution during processing. The Raman
spectroscopy is a spectroscopic method, which depends
on the inelastic scattering of monochromatic light, to
discern vibrational, rotational, and other low-frequency
modes in an object of interest (Fevotte, 2007). The interac-
tion of laser light with molecular vibrations, phonons or
other excitations and subsequent shifting up or down of
laser photons can provide insight about the vibrational
modes. Therefore, this method enables identification of
the textural attributes of a sample due to its remarkable
textural specificity in samples by measuring the molecular
vibrations. In the microwave resonance technology, the
electromagnetic waves with frequencies between
300 MHz and 300 GHz and wave length ranging from
1.0 mm to 1.0 m are commonly applied. The interaction
of the microwave field with solid materials and water mol-
ecules lead to shift in the resonance frequency and attenu-
ation in the resonance bandwidth, which are the basic
infrastructures behind the microwave resonance technique
 M. Aghbashlo et al. / Trends in Food Science & Technology 39 (2014) 76e84 81

(Buschm€ uller, Wiedey, D€oscher, Dressler, & Breitkreutz,
2008). The X-ray densitometry, which is a special type of
the nuclear densitometry technique, uses the attenuation
of photons generated by a radiation source to identify the
density of materials, volume fraction of solids, and most
possibly textural properties of materials through processing
facilities (Wormsbecker, Pugsley, Ommen, Nijenhuis, &
Mudde, 2009). However, most of these techniques need so-
phisticated experimental devices and software for collect-
ing and interpretation of non-specific data. Nevertheless,
the computer vision technology may give a better under-
standing of the texture variations in food processing opera-
tions and also improve the control of processes with
precision and accuracy, being equivalent to the
sophisticated and expensive invasive instruments. It should
be noted that the concept of texture in food industry and
computer vision technology is inherently different from
each other (Zheng, Sun, & Zheng, 2006). Nevertheless, a
good correlation between the image texture with the
texture-related mechanical properties and the cell density
and size has been reported by Gao and Tan (1996a,
1996b) and Sampson, Chang, Vasantha Rupasinghe and
Zaman (2014). However, further studies should be devoted
to demonstrate this hypothesis for more food products un-
derlying different unit operation along with the application
of innovative image acquisition systems and advanced im-
age texture analysis algorithms. As well, Fernandez et al.
(2005) and Quevedo, Jaramillo, Diaz, Pedreschi, and

Start

Capture the RGB color

images from different views

Convert to grey scale image Convert to the desired

Filtering
color space

Masking Thresholding Masking

Labeling the samples

Apply Co-occurrence Apply Fourier Apply wavelet

Matrix method transform transform

Calculate textural Calculate geometrical and Calculate color

parameters shape parameters parameters

Reduce the number of

Main objectives Neural Network
features using PCA method

Make an objective function Fuzzification using FCM Moisture Ratio

and Drying rate

Genetic Algorithm Fuzzy Rules and fuzzy outputs

Defuzzification

Apply the obtained control

values to the process

Wait until the specified time

No
Moisture Content reached
to the desired value?

Yes

Finish

Fig. 4. Flowchart for simultaneous in-line monitoring and controlling of foodstuffs shape, size, color, and texture changes by a computer vision system.
82 M. Aghbashlo et al. / Trends in Food Science & Technology 39 (2014) 76e84
Aguilera (2009) found a clear trend for image texture pa-
rameters variations with drying time. It could be concluded
that the image texture of food profoundly influences by dry-
ing condition and material type, which in turn has a sub-
stantial effect on their mechanical attributes. This also
means that the visual texture monitoring by computer
vision technology provides a possible solution to control
the mechanical texture. Consequently, there is still require-
ment for further investigation and adaptation of the image
texture analysis for improved foodstuffs process monitoring
and control. In general, continuous collecting of the texture
information during food processing using the in-line com-
puter vision system can permit real-time process manipula-
tion to preserve the product characteristics within
predetermined ranges, thereby minimizing the product
loss. Additionally, substantial improvements in the image
acquisition instruments as well as illuminants in conjunc-
tion with advances in the image processing algorithms
render the computer vision systems even more powerful
approach for in-line monitoring and controlling of the
food textural attributes.
Conclusion and future trends
In-line computer vision is a rapid, non-destructive, non-
expansive, efficient, repeatable, precise, and consistent
technique that can be used for in-process analysis of the
organoleptic and sensorial attributes of foodstuffs being
processed through various facilities. It is also noteworthy
that there is a large amount of unexploited potential in
the image data captured during various food processing op-
erations. Desirable advances to manifest this information
are to develop more informative feature extraction algo-
rithms or novel pattern recognition procedures. However,
this method often requires a great extent of computational
efforts for data processing. Such a limitation would be
minimized with further advancement in the computer tech-
nology. As well, the incorrect illumination, improper obser-
vation angle, imperfections or impurity on the processing
media, background light and photo-bleaching can dramati-
cally impress the light intensity and subsequent image qual-
ity. On the other hand, the moisture content of product
surface can modify the reflective characteristics of the ma-
terials and, accordingly, the quality of acquired image.
These problems can be solved by using the advanced imag-
ing and data processing techniques such as confocal laser
scanning microscopy, topographic image processing, and
photometric stereo imaging, particularly for texture mea-
surement. Due to the well-documented ability of in-
process computer vision technology to monitor the shape,
size, color, and texture and the availability of high quality
image acquisition systems with advanced image processing
techniques, pattern recognition, modeling, and optimization
algorithms, it is of interest to survey whether this approach
can be employed for in-line monitoring and controlling of
the foodstuffs quality during drying process. Fig. 4 shows
an innovative flowchart for this purpose. Obviously, the
results achieved by integrated shape, size, color, and texture
information are more precise and complete than the single
measurement. Therefore, it is believed that by extracting
various insights from captured image and assimilating
this information via an appropriate approach, an explicit
relationship between the image data and the product quality
can be established. Generally, the efforts in establishing
real-time image-based quality monitoring tools for food in-
dustry should strive for rapid, reliable, and easily applicable
approaches with feasibilities of smart and informative im-
age feature extraction and feedback control procedures
for process full automation. Thus, computer vision technol-
ogy can be foreseen to be implemented for food dryers in
near future.
Because computer vision systems cannot get complete
information on the food quality indicators such as aroma
and taste, combination of different approaches may provide
a robust method to real-time monitoring of the dryers. In
this regards, richer information is obtained to augment
the capability of the system. Thus, computer vision coupled
with innovative instruments such as electronic tongue and
nose would probably show an effective method to food
quality authentication in drying technology (Ghasemi-
Varnamkhasti, Mohtasebi, & Siadat, 2010). More recently,
innovative and emerging type of electronic nose and tongue
i.e. bioelectronic nose and bioelectronic tongue have been
introduced in which biosensors as sensing elements are em-
ployed (Ghasemi-Varnamkhasti et al., 2011). Such devices
combined with computer vision can be addressed for real-
time food quality assurance during drying process to pro-
vide more effective information on the food quality indica-
tors (taste, odor, and color). Data fusion as an advanced
approach can be considered to get more complete and accu-
rate information compared with individual sources of data
(Shafiee, Minaei, Moghaddam, Ghasemi-Varnamkhasti, &
Barzegar, 2013). However, advanced chemometric tools
should be included in the system to analyze the data gath-
ered. Such idea might be practical in food dryers in close
future.
Acknowledgment
Authors would like to extend their appreciations for
financial support provided by the University of Tehran.
References
Aghbashlo, M., Kianmehr, M. H., & Hassan-Beygi, S. R. (2010).
Drying and rehydration characteristics of sour cherry (Prunus
cerasus L.). Journal of Food Processing and Preservation, 34(3),
351e365.
Aghbashlo, M., Mobli, H., Rafiee, S., & Madadlou, A. (2013). A
review on exergy analysis of drying processes and systems.
Renewable and Sustainable Energy Reviews, 22, 1e22.
Bonazzi, C., & Dumoulin, E. (2011). Quality changes in food
materials as influenced by drying processes. In E. Tsotsas, &
A. S. Mujumdar (Eds.), Modern drying technology, volume 3:
Product quality and formulation (pp. 1e20). Weinheim, Germany:
Wiley-VCH.
 M. Aghbashlo et al. / Trends in Food Science & Technology 39 (2014) 76e84
Burggraeve, A., Monteyne, T., Vervaet, C., Remon, J. P., & Beer, T. D.
(2012). Process analytical tools for monitoring, understanding and
control of pharmaceutical fluidized bed granulation: a review.
European Journal of Pharmaceutics and Biopharmaceutics, 83(1),
2e15.
Buschm€ uller, C., Wiedey, W., D€oscher, C., Dressler, J., & Breitkreutz, J.
(2008). In-line monitoring of granule moisture in fluidized-bed
dryers using microwave resonance technology. European Journal of
Pharmaceutics and Biopharmaceutics, 69(1), 380e387.
Campos-Mendiola, R., Hernandez-Sanchez, H., Chanona-Perez, J. J.,
Alamilla-Beltran, L., Jimenez-Aparicio, A., Fito, P., et al. (2007).
Non-isotropic shrinkage and interfaces during convective drying
of potato slabs within the frame of the systematic approach to food
engineering systems (SAFES) methodology. Journal of Food
Engineering, 83(2), 285e292.
Chen, Y., & Martynenko, A. (2013). Computer vision for real-time
measurements of shrinkage and color changes in blueberry
convective drying. Drying Technology, 31(10), 1114e1123.
Chen, L., & Opara, U. L. (2013). Texture measurement approaches in
fresh and processed foods e a review. Food Research
International, 51(2), 823e835.
Dowlati, M., Guardia, M. D. L., & Mohtasebi, S. S. (2012).
Application of machine-vision techniques to fish-quality
assessment. TrAC Trends in Analytical Chemistry, 40, 168e179.
FDA. (2004). Guidance for industry; PATeA framework for innovative
pharmaceutical development, manufacturing and quality
assurance, US Department of Health and Human Services.
Fernandez, L., Castillero, C., & Aguilera, J. M. (2005). An application
of image analysis to dehydration of apple discs. Journal of Food
Engineering, 67(1), 185e193.
Fevotte, G. (2007). In situ Raman spectroscopy for in-line control of
pharmaceutical crystallization and solids elaboration processes: a
review. Chemical Engineering Research and Design, 85(7),
906e920.
Figura, L. O., & Teixeira, A. A. (2007). Optical properties. In
L. O. Figura, & A. A. Teixeira (Eds.), Food physics: Physical
propertiesdMeasurement and applications (pp. 391e416). Berlin:
Springer.
Gao, X., & Tan, J. (1996a). Analysis of expanded-food texture by
image processing part I: geometric properties. Journal of Food
Process Engineering, 19(4), 425e444.
Gao, X., & Tan, J. (1996b). Analysis of expanded-food texture by
image processing part II: mechanical properties. Journal of Food
Process Engineering, 19(4), 445e456.
Ghasemi-Varnamkhasti, M., Mohtasebi, S. S., Rodriguez-
Mendez, M. L., Siadat, M., Ahmadi, H., et al. (2011). Electronic
and bioelectronic tongues, two promising analytical tools for the
quality evaluation of non alcoholic beer. Trends in Food Science &
Technology, 22(5), 245e248.
Ghasemi-Varnamkhasti, M., Mohtasebi, S. S., & Siadat, M. (2010).
Biomimetic-based odor and taste sensing systems to food quality and
safety characterization: an overview on basic principles and recent
achievements. Journal of Food Engineering, 100(3), 377e387.
Ghiaus, A. G., Margaris, D. P., & Papanikas, D. G. (1997).
Mathematical modeling of the convective drying of fruits and
vegetables. Journal of Food Science, 62(6), 1154e1157.
Hosseinpour, S., Rafiee, S., & Mohtasebi, S. S. (2011). Application of
image processing to analyze shrinkage and shape changes of
shrimp batch during drying. Drying Technology, 29(12),
1416e1438.
Hosseinpour, S., Rafiee, S., Mohtasebi, S. S., & Aghbashlo, M. (2013).
Application of computer vision technique for on-line monitoring
of shrimp color changes during drying. Journal of Food
Engineering, 115(1), 99e114.
Jinorose, M., Prachayawarakorn, S., & Soponronnarit, S. (2010).
Development of a computer vision system and novel evaluation
83
criteria to characterize color and appearance of rice. Drying
Technology, 28(9), 1118e1124.
Kerr, W. L. (2007). Food drying and evaporation processing
operations. In M. Kutz (Ed.), Handbook of farm, dairy, and food
machinery (pp. 303e340). New York: William Andrew Publishing.
Koç, B., Eren, I., & Kaymak Ertekin, F. (2008). Modelling bulk density,
porosity and shrinkage of quince during drying: the effect of drying
method. Journal of Food Engineering, 85(3), 340e349.
Krokida, M. K., & Maroulis, Z. B. (1997). Effect of drying method on
shrinkage and porosity. Drying Technology, 15(10), 2441e2458.
Lozano, J. E., Rotstein, E., & Urbicain, M. J. (1983). Shrinkage,
porosity and bulk density of foodstuffs at changing moisture
contents. Journal of Food Science, 48(5), 1497e1502.
Madamba, P. S., Driscoll, R. H., & Buckle, K. A. (1994). Bulk density,
porosity and resistance to airflow of garlic slices. Drying
Technology, 12(4), 937e954.
Maskan, A., Kaya, S., & Maskan, M. (2002). Effect of concentration
and drying processes on color change of grape juice and leather
(pestil). Journal of Food Engineering, 54(1), 75e80.
Mayor, L., Moreira, R., & Sereno, A. M. (2011). Shrinkage, density,
porosity and shape changes during dehydration of pumpkin
(Cucurbita pepo L.) fruits. Journal of Food Engineering, 103, 29e37.
Mayor, L., & Sereno, A. M. (2004). Modelling shrinkage during
convective drying of food materials: a review. Journal of Food
Engineering, 61(3), 373e386.
Mendoza, F., & Aguilera, J. M. (2004). Application of image analysis
for classification of ripening bananas. Journal of Food Science,
69(9), 471e477.
Michailidis, P. A., Krokida, M. K., Bisharat, G. I., Marinos-Kouris, D.,
& Rahman, M. S. (2010). Measurement of density, shrinkage, and
porosity. In M. S. Rahman (Ed.), Food properties handbook (pp.
397e416). New York: CRC Press.
Mohebbi, M., Akbarzadeh-T, M. R., Shahidi, F., Moussavi, M., &
Ghoddusi, H. B. (2009). Computer vision system (CVS) for
moisture content estimation in dehydrated shrimp. Computers and
Electronics in Agriculture, 69(2), 128e134.
Mulet, A., Garcia-Reverter, J., Bon, J., & Berna, A. (2000). Effect of
shape on potato and cauliflower shrinkage during drying. Drying
Technology, 18(6), 1201e1219.
Nadian, M. H., Rafiee, S., Aghbashlo, M., Hosseinpour, S., &
Mohtasebi, S. S. (2014). Continuous real-time monitoring and
neural network modeling of apple slices color changes during hot
air drying. Food and Bioproducts Processing, . http://dx.doi.org/
10.1016/j.fbp.2014.03.005.
Phadungath, C. (2003). Basic measurement for food texture. Lecturer
in Food Science and Technology Program, Faculty of Science and
Technology, Surindra Rajabhat University.
Quevedo, R., Jaramillo, M., Diaz, O., Pedreschi, F., & Aguilera, J. M.
(2009). Quantification of enzymatic browning in apple slices
applying the fractal texture Fourier image. Journal of Food
Engineering, 95(2), 285e290.
Rahman, M. S., Perera, C. O., Chen, X. D., Driscoll, R. H., &
Potluri, P. L. (1996). Density, shrinkage and porosity of calamari
mantle meat during air drying in a cabinet dryer as a function of
water content. Journal of Food Engineering, 30(1), 135e145.
Ramos, I. N., Silva, C. L. M., Sereno, A. M., & Aguilera, B. J. M. (2004).
Quantification of microstructural changes during first stage air
drying of grape tissue. Journal of Food Engineering, 62, 159e164.
Sampson, D. J., Chang, Y. K., Vasantha Rupasinghe, H. P., &
Zaman, Q. U. (2014). A dual-view computer-vision system for
volume and image texture analysis in multiple apple slices drying.
Journal of Food Engineering, 127, 49e57.
Shafiee, S., Minaei, S., Moghaddam, N., Ghasemi-Varnamkhasti, M., &
Barzegar, M. (2013). Potential application of machine vision to honey
characterization. Trends in Food Science & Technology, 30(2),
174e177.
84 M. Aghbashlo et al. / Trends in Food Science & Technology 39 (2014) 76e84
da Silva, C. A. M., Butzge, J. J., Nitz, M., & Taranto, O. P. (2014).
Monitoring and control of coating and granulation processes in
fluidized bedsea review. Advanced Powder Technology, 25(1),
195e210.
Watzenig, D., & Fox, C. (2009). A review of statistical modelling and
inference for electrical capacitance tomography. Measurement
Science and Technology, 20(5), 052002.
Whitaker, M., Baker, G. R., Westrup, J., Goulding, P. A., Rudd, D. R.,
Belchamber, R. M., et al. (2000). Application of acoustic emission
to the monitoring and end point determination of a high shear
granulation process. International Journal of Pharmaceutics,
205(1), 79e91.
Wormsbecker, M., Pugsley, T., Ommen, J. R. V., Nijenhuis, J., &
Mudde, R. (2009). Effect of distributor design on the bottom zone
hydrodynamics in a fluidized bed dryer using 1-D X-ray
densitometry imaging. Industrial & Engineering Chemistry
Research, 48(15), 7004e7015.
Yadollahinia, A., & Jahangiri, M. (2009). Shrinkage of potato slice
during drying. Journal of Food Engineering, 94(1), 52e58.
Yan, Z., Sousa-Gallagher, J. M., & Oliveira, A. R. F. (2008). Shrinkage
and porosity of banana, pineapple and mango slices during air-
drying. Journal of Food Engineering, 83(3), 430e440.
Zheng, C., Sun, D. W., & Zheng, L. (2006). Recent applications of
image texture for evaluation of food qualitiesea review. Trends in
Food Science & Technology, 17(3), 113e128.
Zogzas, N. P., Maroulis, Z. B., & Marinos-Kouris, D. (1994). Densities,
shrinkage and porosity of some vegetables during air drying.
Drying Technology, 12(7), 1653e1666.