Journal of Cereal Science 74 (2017) 194e199

Contents lists available at ScienceDirect

Journal of Cereal Science

journal homepage: www.elsevier.com/locate/jcs

Application of image analysis to optimization of the bread-making

process based on the acceptability of the crust color
Wilson Castro a, *, Jimy Oblitas a, Tony Chuquizuta b, Himer Avila-George c
a
School of Engineering, Universidad Privada del Norte, Vía de evitamiento S/N. Cuadra 15, Cajamarca, Cajamarca, Peru
b
School of Engineering and Agricultural Science, Universidad Nacional Toribio Rodríguez de Mendoza de Amazonas, Pampas de Higos Urco S/N, Ciudad
Universitaria, Chachapoyas, Amazonas, Peru
c
CONACYT e CICESE, Unidad de Transferencia Tecnolo
gica Tepic, Andador 10 #109, entre calles 3 y 4, Ciudad del Conocimiento, Tepic, Nayarit, Mexico

a r t i c l e i n f o a b s t r a c t

Article history: Consumption of bread and the demands concerning its quality features, being one of them its appear-
Received 13 October 2016 ance, have been experiencing rapid growth. Thus, the standardization of its production aiming to keep its
Received in revised form quality, applying new methods. The objective of this research was to develop a method to optimize the
26 January 2017
bread-making processes based on the acceptability of its crust color. For this effect, bread was experi-
Accepted 2 February 2017
Available online 7 February 2017
mentally produced using a Box-Behnken experimental design with three factors (sugar-flour relation,
Baking temperature and Baking time) and three answer variables (L*, a*, b* ¼ parameters of CIELab color
space); determination of color, by means of the acquisition, pre-processing, and analysis of images of
Keywords:
Image analysis
bread samples until getting the color expressed in CIELab coordinates; an analysis of sensorial acceptance
CIELab was made determining the L*, a*, and b* with the highest acceptance by consumers; finally, the opti-
Darkening mization of the production process was made based on the L*, a*, and b* parameters, getting the optimal
Surface of response production parameters. The results show that by using the proposed method, it is possible to correlate
the parameters of CIELab color space and the acceptance of the final consumer aiming to optimize bread
making processes, it means getting bread with crust color of maximum acceptability.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction properties of the materials and conditions of functioning, is one of

Cereals and its derivatives, especially bread, constitute an
important element of the human diet from the earliest civilizations,
and its consumption is experiencing a sharp increase accompanied
by more strict demands in terms of its appearance, texture, and
flavor (Abdullah, 2008; Gon ~ i et al., 2008; Purlis, 2011; Purlis and
Salvadori, 2007). Appearance stands out among the aforemen-
tioned features since it's closely linked to the decision of buying
(Purlis and Salvadori, 2007), constituting and important parameter
to study in order to improve its quality and standardize its pro-
duction process (Abdullah, 2008).
Control of the bread-making process requires to manipulate a
series of parameters and is based on the criteria of experts, which
makes it harder to standardize production quality (Lara et al., 2011).
Then, to remove the subjective element of quality control activities
and manage to control the time of the process, on the basis of the
* Corresponding author.
E-mail address: wilson.castro@upn.edu.pe (W. Castro).
the main concerns of food engineers (Gon ~ i et al., 2008); being
essential to take into account the phenomena of transportation and
its effects during the productive process (Purlis, 2011, 2012). Thus,
to determine, predict, and control the development of color of the
bread crust based on production parameters, play a vital role in the
bread-making industry (Du et al., 2012).
In the bread-making process, the stage of baking is the last and
most important one due to physiochemical and organoleptic
changes in the bread dough; giving this dough typical bread char-
acteristics such as: texture, volume, formation of crumbs and at
last, its most important characteristic, formation and coloring of the
crust (Mohd et al., 2009; Paquet-Durand et al., 2012; Purlis, 2012;
Purlis and Salvadori, 2007). The development of color happens over
the last stages of the baking process as a result of a complex process
of biochemical reactions and colored compounds that these
generate (Abdullah, 2008; Ahrne
et al., 2007; Broyart et al., 1998).
These reactions were first observed by the French chemist Louis
Camille Maillard in 1912 and were lately named Maillard reaction
(Mundt and Wedzicha, 2007; Purlis, 2012).
Maillard reactions include all those which imply compounds

http://dx.doi.org/10.1016/j.jcs.2017.02.002
0733-5210/© 2017 Elsevier Ltd. All rights reserved.
 W. Castro et al. / Journal of Cereal Science 74 (2017) 194e199 195

with amino groups and carbonyl groups present in foods including for bread known as traditional type (a typical food of mountain
amines, amino acids, and proteins which interact with sugars, al- region in Peru). The equipment used to bread making is composed
dehydes, and ketones, as well as products of lipids oxidation (Eskin of: a stand mixer-kneader model KN-25-NOVA, an electronic scale
et al., 2013; Purlis, 2012). The speed of these reactions depends on model AG/DSW-40, a divider and a bread-making oven model
the temperature, pH, water activity (aw), the absence of metallic 2000-NOVA. Considering the weight of flour as 100%, the in-
cations, and the conformation and structure of the sugar involved gredients and its percentage, to produce the bread samples were:
(Abdullah, 2008; Mundt and Wedzicha, 2007; Walker et al., 2012). flour (100%), water (62%), dried yeast (5%), vegetable fat (3%), salt
Monitoring of the formation of these compounds in bread-making (1.8%) and sugar [0.8e1.2%]. The sugar/flour relation (SFR) is one the
products is being done using traditional chemical methods: color- study variables.
imetry, chromatographic separation, measurement of the genera- The bread-making production followed the straight dough
tion of carbon dioxide, ultraviolet and infrared analysis (IR method commented in the review of Mondal and Datta (2008).
spectrum), which are useful to evaluate final products (Purlis and After mixing the ingredients described above, eight steps were
Salvadori, 2007) and inappropriate to measure changes of color followed to prepare the bread samples: (1) first weighing, (2)
during the baking process. mixing-kneading, (3) second weighing, (4) division, (5) balling, (6)
This need for more exhaustive analysis, less subjective, non- fermentation, (7) baking, and (8) cooling. Finally, units of 35 g were
destructive, and applicable to the production line, has led to the obtained, which were used to develop the desirability analysis, the
development of techniques according to those requirements, such as image analysis and correlation between both analyzes.
the technique of computer vision (SVC) upon foods (Leo
n et al., 2006).
This technique simulates the mode of capturing and interpreting
2.1.2. System for image acquisition during the baking process
information of the human eye (Oblitas et al., 2016) and has generated
The system for image acquisition used in this research is shown
a revolution in the processes for the evaluation of food quality (Sun,
in Fig. 1, its design details are: (1) The walls have a dimension of
2004), allowing to measure the physical properties of food at
50  40 cm, and inside are matte black; (2) it uses three light bulbs
microscopic or macroscopic levels (Abdullah, 2008). The SVC tech-
daylight type, power 18 w, and equilateral location (10 cm edge);
nique allows to quantify color in real time (Broyart et al., 1998) and to
(3) A chamber with an image resolution of 14.5 megapixels, an
correlate its characteristics with the acceptability of the consumer or
image size of 4320  3240 and equipped with a CCD sensor; and (4)
production parameters (Brosnan and Sun, 2004); it also has advan-
a computer with Intel Core i5 inside and 4 GB of RAM. The samples
tages over conventional colorimetric methods since it analyzes the
extracted during the bread-making process were subjected to im-
whole surface and quantifies morphologic characteristics such as
age acquisition in the *.jpg format (Joint Photographic Experts
cracks, protuberances, and other defects (Lara et al., 2011).
Group).
According to Mundt and Wedzicha (2007), colorimetry is the
most elemental analysis technique of a food, and it provides in-
formation about the color of a product representing it in RGB values 2.2. Methodology
(reflectance of the sample when the intensity of the reflected light
is analyzed through standard red, green, and blue filters) or The methodology proposed for this research is shown in Fig. 2:
expressing them in tristimulus values according to the XYZ, HUV or
CIELab color spaces (Lara et al., 2011; Mundt and Wedzicha, 2007). 2.2.1. Experimental production
Space most often used for instrumental control within the industry The pilot production was made using an experimental design of
is the CIELab or Lab, since it defines consistent colors and in an the type of response surface, Box-Behnken with central points and
independently from the device used to acquire the images and no repetition. We considered as factors the process parameters at
whose parameters are (Leo
n et al., 2006; Wu and Sun, 2013):
three levels (low, medium, high), and as response variables the
parameters of color in the CIELab space (L*, a*, b*). The process
- L* or Luminance [0 to 100]; parameters levels are:
- a* or coloring from green to red [-120 to120];
- b* or coloring from blue to yellow [-120 to 120];  Sugar/Flour relation: 0.8, 1 and 1.2
 Baking temperature ( C): 160, 165 and 170
CIELab space has been used in research papers such as in the  Baking time (min): 16, 18 and 20
determination of russeting of mango (Zheng and Lu, 2012), the
presence of anthocyanins in grapes (Liang et al., 2011), and con-
cerning baked products in the characterization of crackers, (Lara
et al., 2011), Italian bread, (Brescia et al., 2007), integral bread
(Flander et al., 2007), steam rolls (Ma et al., 2014), among others.
Then, considering the development of new techniques to eval-
uate color on foods and the importance of the bread crust color, the
present research has been done in order to develop a new method
to optimize bread-making processes based on the acceptability of
the crust color, setting as objectives: (a) experimental bread pro-
duction (b) determination of color in CIELab space (c) analysis of
acceptability, and (d) optimization of the production process.

2. Materials and methodology

2.1. Materials

2.1.1. Bread samples

The bread samples were elaborated using the standard recipe Fig. 1. System for image acquisition.
196 W. Castro et al. / Journal of Cereal Science 74 (2017) 194e199

2 3 2 3 2 3 2 3
x r 0:4124 0:3576 0:1805 r
4 y 5 ¼ 100*½M*4 g 5 ¼ 100*4 0:2126 0:7152 0:0722 5*4 g 5
z b 0:0193 0:1192 0:9505 b
(4)
From the x, y, z values, the X, Y, Z values were determined by
means of equation (5)

Fig. 2. Methodology for the optimization of the parameters of the process.
The number of required samples and the levels per factor in
each sample were determined using the software Statgraphics
Centurion VI.
2.2.2. Determination of color parameters
Based on Sun's proposal (2004), a logical sequence was devel-
oped, see Fig. 2. Said sequence is necessary to acquire, pre-process
and determine the color of bread samples, by means of image
analysis.
Upon analyzing the images of the bread samples, the regions of
interest (ROIs) were determined, which contain the pixels consti-
tuting the sample; the RGB information of the ROI's was extracted,
from which the color in CIELab coordinates was calculated.
The logical sequence was implemented in the Matlab v. 2015a
mathematical software. In this logical sequence, the color,
expressed in CIELab space, was determined using (1)e(11) equa-
tions according to proposals of Leo
n et al. (2006); Saldan
~ a et al.
(2014), said equations are shown in detail below:
Step a. Getting the RGB normalized values.

2:4
R R þ 0:055
if > 0:04045 /r ¼ ; else r ¼
255 1:055 12:92


G G þ 0:055 2:4 G
if > 0:04045 /g ¼ ; else g ¼
255 1:055 12:92


B B þ 0:055 2:4 B et al. (2016).
if > 0:04045 /b ¼ ; else b ¼
255 1:055 12:92
Step b. Getting the XYZ parameters.
RGB normalized values were transformed into xyz values upon
multiplying these by the M matrix of the set illuminant e observer.
D65-2 was used since it is considered standard by the Commission
Internationale d’Eclairage (CIE).
x y z
X¼ ;Y ¼ ;Z ¼ (5)
Xn Xn Zn
Values Xn, Yn and Zn, are 95.047, 100 and 108.883 respectively
and correspond to tristimulus values gotten by means of the
weighted average method of medium coordinates (Dl ¼ 1 nm),
using the CIE 1964 standard observer (2 of the visual area) and D65
CIE standard illuminant.
Step c. Determination of color parameters in CIELab space.
In this step, variables var_X, var_Y and var_Z were determined
by means of equations from 6 to 8


1 16
if X > 0:008856 /var X ¼ ðXÞ3 ; else var X ¼ ð7:787*XÞ þ
116
(6)


1 16
if Y > 0:008856 /var Y ¼ ðYÞ3 ; else var Y ¼ ð7:787*YÞ þ
116
(7)


1 16
if Z > 0:008856 /var Z ¼ ðZÞ3 ; else var Z ¼ ð7:787*ZÞ þ
116
(8)
Finally, L*, a* and b* color variables were calculated by means of
equations from 9 to 11.
L* ¼ ð116*var YÞ  16 (9)
a* ¼ 500*ðvar X  var YÞ (10)
b* ¼ 200*ðvar Y  var ZÞ (11)
2.2.3. Analysis of sensorial acceptability
The sensorial acceptability, for each CIELab parameter, was
determined according to their preference, expressed in percentages
similarly to Garcia-Díez et al. (2016). The sensory evaluation was
carried out on bread samples arranged in three groups. Each group
consisted of three samples with minimum, intermediate and
maximum values of L*, a*, or b* randomly organized.
R The panel consisted of 45 untrained judges (aged between 21
(1)
and 43 years), the number of judges was selected following the
recommendations reported in Hough et al. (2006). Judges evalu-
ated visually samples, selecting in each group the samples (CIELab
(2) parameter value) that they preferred. For each CIELab parameter,
the sensorial acceptability was determined according to its pref-
erence, expressed it in percentage manner, similar to García-Díez
(3)
2.2.4. Optimization of the production process
For the optimization, as objective value for each variable
response, we set the one with the highest acceptability gotten in
the previous step, keeping the factors within the range of work. The
optimal combination of factors was combined into a general
 W. Castro et al. / Journal of Cereal Science 74 (2017) 194e199
compound function, D(x), named Desirability Function, see equa-
tion (12) (Mudgil et al., 2016).
DðXÞ ¼ fd1 :d2 …dn g1=n
In which:
D ¼ General desirability
d ¼ Desirability per variable response
n ¼ Number of variable responses
Numeric optimization determined a point in which the conve-
nience function improves to its maximum level, and its graphics of
response surface were generated with the help of the statistical
package Statgraphics Centurion VI.
3. Results and discussions
3.1. Images of the baking process
Fig. 3 shows different RGB images gotten during the baking
process according to the time and temperature scales of the
experimental design.
Images shown in Fig. 3 show uniform illumination and enables
to extract properly the color information from the breast crust;
other researchers have used images at the end of the production
process to determine the effect of the treatment of superficial color
of baked products, such as Abdullah (2008) for Muffins, Broyart
et al. (1998) for crackers, Lara et al. (2011) for sponge cakes, or
Fig. 3. Images of samples (a) sfr ¼ 0.8; (b) sfr ¼ 1.0 (c) sfr ¼ 1.2.
197
during the process, like the work of Paquet-Durand et al. (2012) on
the changes of the color on the crust of round bread. However, the
difference between one and the other case is the difficulty in
(12) standardizing the conditions of image acquisition which as Du and
Sun (2004) state, are the first step in the acquisition of high-quality
images, and as Paquet-Durand et al. (2012) reported, it can be
influenced by the changes due to the expansion of dough and the
lenses getting cloudy, in case it is placed inside the oven, or the
window's cloudiness if the camera is placed outside.
3.2. CIELab parameters in the bread crust during baking
The logical sequence (script) was implemented in Matlab 2015a
in order to get the color parameters in the CIELab space. This
sequence is similar to the one proposed by Saldan ~ a et al. (2014).
This sequence used the contrast between sample and background
for the segmentation process as an initial action in the color
measuring process. There are other methods which could be used
for the segmentation, such as artificial intelligence techniques
proposed by Du and Sun (2004). However, due to the simplicity of
the method used in this research, segmentation by thresholding.
On the other hand, in order to determine the CIELab parameters,

n et al. (2006) equations were used, which could have been
Leo
done correlating values obtained by means of a colorimeter and the
RGB of the image, this would leave the acquisition conditions out.
The behavior of parameters L*, a*, and b*, gotten upon analyzing the
samples, is shown in Fig. 4.
The most used and reported color parameter to analyze color of
the bread crust in CIELab coordinates is L*, for its relation with the
198 W. Castro et al. / Journal of Cereal Science 74 (2017) 194e199

brilliance of the crust (Broyart et al., 1998; Lara et al., 2011; Purlis
and Salvadori, 2009); likewise, color indexes have been reported
such as intensity in the work of Paquet-Durand et al. (2012), or total
(E*) color change in function of L*, a*, and b* Purlis and Salvadori
(2007). The experimental values were: Temperature [160e170]
 C, Time [16e20] min, L* [60.5e67.4], a* [4.5e1.5], and b*
[22.6e45.9].
The experimental values are basically similar to those reported
by Belghith et al. (2016), while they differ from those presented by
Siddiq et al. (2009) and Ma et al. (2014), this due to the different
composition or type of production process. Likewise, Purlis and
Salvadori (2009), widely recognized authors report solely the
value L*, upon considering it of higher relevance for modelling of
processes of russeting, fluctuating in a range which includes values
reported in this research.
3.3. Acceptability of bread samples
The color parameters were visually evaluated, determining the
degree of acceptability between minimum, maximum, and inter-
mediate levels which the process produces according to the con-
ditions of the operation. The values of maximum acceptability for
CIELab parameters were L* ¼ 62.9, a* ¼ 1.5, and b* ¼ 38.7.
Results show that the consumer favored intermediate L* values,
it means a soft shade of brown, and based on the combination of
parameters a* and b*, light brown colorings.
(jDa*j ¼ 0:67 - 43%); it means that it was expected that the product

ndez (1993) explains, this
gets slightly less reddish shades. As Herna
difference is due both to the first-order sensitivity of the objective
function (desirability) and to the values of the selected design
variables. The same author mentions that in order to improve the
sensitivity it is necessary to consider the suitability of modifying
some of the parameters of production. However, since the bread
making process used is based on a recipe described in section 2.1.1,
the values of any of the production parameters can not be modified.
Therefore, it could be considered to increase the number of points
in the design or, alternatively, to use objective functions of a
different order as reported in Barman (2016).
The scope of this optimization includes three production pa-
rameters; however, it must be understood that the color of the
obtained product is the result of the production parameters in the
optimization process, and of other parameters such as flour type,
presence of milk, initial and final humidity, among others
(Abdullah, 2008).

3.4. Optimized production parameters

Fig. 5 shows the response surface of the optimization of the

production parameters (sfr, T, t) according to the L*, a*, and b* with
higher acceptability gotten in the previous step. Likewise, Table 1
shows optimized values for the production parameters.
Optimized values of the response variables are close to those
expected, only one difference exists in the color parameter a* Fig. 5. Response surface with the optimized production parameters.

Fig. 4. Respond surfaces of parameters L*, a*, and b*.

 W. Castro et al. / Journal of Cereal Science 74 (2017) 194e199 199

Table 1 Flander, L., Salmenkallio-Marttila, M., Suortti, T., Autio, K., 2007. Optimization of
Optimized production parameters and response variables. ingredients and baking process for improved wholemeal oat bread quality. LWT
- Food Sci. Technol. 40 (5), 860e870.
Factor Units Minimum Maximum Optimized García-Díez, J., Alheiro, J., Pinto, A., Soares, L., Falco, V., Fraqueza, M., Patarata, L.,
 2016. Behaviour of food-borne pathogens on dry cured sausage manufactured
Temperature C 160 170 170.0
with herbs and spices essential oils and their sensorial acceptability. Food
Sugar-flour relation % 0.8 1.2 1.1
Control. 59, 262e270. http://dx.doi.org/10.1016/j.foodcont.2015.05.027.
Time minutes 16 20 17.5 Gon~ i, S., Oddone, S., Segura, J., Mascheroni, R., Salvadori, V., 2008. Prediction of
Response variable Minimum Maximum Expected Optimized foods freezing and thawing times: artificial neural networks and genetic al-
gorithm approach. J. Food Eng. 84 (1), 164e178.
L* 60.47 67.4 62.91 62.91 Herna
ndez, S., 1993. Del disen ~ o convencional al disen ~o o

ptimo. Posibilidades y
a* 4.48 1.54 1.54 0.87 Variantes. Parte II. Optimizacio
n multiobjetivo y sensibilidad de la solucio
n
B* 22.6 45.99 38.72 38.72
ptima. Rev. Int. me
o
todos nume
ricos ca
lculo disen
~ o Ing. 9 (3), 259e270.
Hough, G., Wakeling, I., Mucci, A., Chambers IV, E., Gallardo, I., Alves, L., 2006.
Value of the bread crust color parameters obtained by optimizing the production Number of consumers necessary for sensory acceptability tests. Food Qual.
parameters are indicated in bold. Prefer. 17 (6), 522e526. http://dx.doi.org/10.1016/j.foodqual.2005.07.002.
Lara, E., Corte
s, P., Briones, V., Pe
rez, M., 2011. Structural and physical modifications
of corn biscuits during baking process. LWT - Food Sci. Technol. 44 (3),
622e630.
4. Conclusions

n, K., Mery, D., Pedreschi, F., Leo
Leo
n, J., 2006. Color measurement in L*a*b* units
from RGB digital images. Food Res. Int. 39 (10), 1084e1091.
This study successfully developed and applied a methodology Liang, Z., Sang, M., Fan, P., Wu, B., Wang, L., Yang, S., Li, S., 2011. CIELAB coordinates
for the optimization of three bread production parameters: sugar- in response to berry skin anthocyanins and their composition in Vitis. J. Food
Sci. 76 (3), 490e497.
flour relation (%), temperature ( C) and baking time (minutes). Ma, S., Wang, X., Zheng, X., Tian, S., Liu, C., Li, L., Ding, Y., 2014. Improvement of the
This methodology required the production of samples based on an quality of steamed bread by supplementation of wheat germ from milling
experimental design of the type of response surface, the analysis of process. J. Cereal Sci. 60 (3), 589e594.
Mohd, Y., Chin, N., Yusof, Y., Abdul, R., 2009. Bread crust thickness measurement
the crust color expressed in CIELab space by means of image using digital imaging and L a b colour system. J. Food Eng. 94 (3e4), 366e371.
analysis and its sensorial evaluation in order to determine the de- Mondal, A., Datta, A., 2008. Bread baking e a review. J. Food Eng. 86 (4), 465e474.
gree of acceptability at the different levels of L*, a*, and b*. Through http://dx.doi.org/10.1016/j.jfoodeng.2007.11.014.
Mudgil, D., Barak, S., Khatkar, B., 2016. Optimization of bread firmness, specific loaf
an optimization process of multiple parameters, it was concluded volume and sensory acceptability of bread with soluble fiber and different
that it is possible to determine the optimum production parameters water levels. J. Cereal Sci. 70, 186e191.
and the optimized values of the response variables. Mundt, S., Wedzicha, B., 2007. A kinetic model for browning in the baking of bis-
cuits: effects of water activity and temperature. LWT - Food Sci. Technol. 40 (6),
1078e1082.
References Oblitas, J., Castro, W., Mayor, L., 2016. Effect of different combinations of size and
shape parameters in the percentage error of classification of structural ele-
Abdullah, M., 2008. Quality Evaluation of Bakery Products. En Computer Vision ments in vegetal tissue of the pumpkin Cucurbita pepo L. using probabilistic
Technology for Food Quality Evaluation. Academic Press, Amsterdam, neural networks. Rev. Fac. Ing. 78, 30e37.
pp. 481e522. Paquet-Durand, O., Solle, D., Schirmer, M., Becker, T., Hitzmann, B., 2012. Monitoring
Ahrne
, L., Andersson, C., Floberg, P., Rose
n, J., Lingnert, H., 2007. Effect of crust baking processes of bread rolls by digital image analysis. J. Food Eng. 111 (2),
temperature and water content on acrylamide formation during baking of 425e431.
white bread: steam and falling temperature baking. LWT - Food Sci. Technol. 40 Purlis, E., 2011. Bread baking: technological considerations based on process
(10), 1708e1715. modelling and simulation. J. Food Eng. 103 (1), 92e102.
Barman, G., 2016. Max-min desirability function approach to optimize multiple Purlis, E., 2012. Baking process design based on modelling and simulation: towards
response optimization problems by honey bee colony strategy. J. Comput. Math. optimization of bread baking. Food Control. 27 (1), 45e52.
Sci. 7 (6), 328e335. Purlis, E., Salvadori, V., 2007. Bread browning kinetics during baking. J. Food Eng. 80
Belghith, L., Chaari, F., Maaloul, M., Kallel, F., Abdelkafi, L., Ellouz_Chaabouni, S., (4), 1107e1115.
Ghribi-Aydi, D., 2016. Wheat bread enrichment by pea and broad bean pods Purlis, E., Salvadori, V., 2009. Modelling the browning of bread during baking. Food
fibers: effect on dough rheology and bread quality. LWT - Food Sci. Technol. 73, Res. Int. 42 (7), 865e870.
584e591. Saldan ~ a, E., Siche, R., Castro, W., Huama
n, R., Quevedo, R., 2014. Measurement
Brescia, M., Sacco, D., Sgaramella, A., Pasqualone, A., Simeone, R., Peri, G., Sacco, A., parameter of color on yacon (Smallanthus sonchifolius) slices using a computer
2007. Characterisation of different typical Italian breads by means of traditional, vision system. LWT - Food Sci. Technol. 59 (2, Part 2), 1220e1226.
spectroscopic and image analyses. Food Chem. 104 (1), 429e438. Siddiq, M., Nasir, M., Ravi, R., Butt, M.S., Dolan, K.D., Harte, J.B., 2009. Effect of
Brosnan, T., Sun, D., 2004. Improving quality inspection of food products by com- defatted maize germ flour addition on the physical and sensory quality of
puter visioneea review. J. Food Eng. 61 (1), 3e16. wheat bread. LWT - Food Sci. Technol. 42 (2), 464e470.
Broyart, B., Trystram, G., Duquenoy, A., 1998. Predicting colour kinetics during Sun, D., 2004. Computer visioneean objective, rapid and non-contact quality
cracker baking. J. Food Eng. 35 (3), 351e368. evaluation tool for the food industry. J. Food Eng. 61 (1), 1e2.
Du, C., Cheng, Q., Sun, D., 2012. 16-Computer Vision in the Bakery Industry. En Walker, S., Seetharaman, K., Goldstein, A., 2012. Characterizing physicochemical
Computer Vision Technology in the Food and Beverage Industries. Woodhead changes of cookies baked in a commercial oven. Food Res. Int. 48 (1), 249e256.
Publishing, pp. 422e450. Wu, D., Sun, D., 2013. Colour measurements by computer vision for food quality
Du, C., Sun, D., 2004. Recent developments in the applications of image processing controleA review. Trends Food Sci. Technol. 29 (1), 5e20.
techniques for food quality evaluation. Trends Food Sci. Technol. 15, 230e249. Zheng, H., Lu, H., 2012. A least-squares support vector machine (LS-SVM) based on
Eskin, N., Ho, C., Shahidi, F., 2013. Chapter 6-browning reactions in foods. En fractal analysis and CIELab parameters for the detection of browning degree on
Biochemistry of Foods, third ed. Academic Press, San Diego, pp. 245e289. mango (Mangifera indica L.). Comput. Electron. Agric. 83, 47e51.