Documente Academic
Documente Profesional
Documente Cultură
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/222141091
CITATIONS READS
57 221
3 authors, including:
Christopher A Brown
Worcester Polytechnic Institute
111 PUBLICATIONS 2,056 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
influence of material properties and fabrication parameters on 3d printing materials View project
Industry 4.0 for SMEs - EU Horizon 2020 RISE program View project
All content following this page was uploaded by Christopher A Brown on 02 August 2016.
a
Department of Chemical Engineering and Bioprocesses, Ponticia Universidad Catolica de Chile, P.O. Box 306, Santiago 22, Chile
b
Mechanical Engineering Department, Worcester Polytechnic Institute, Worcester, MA 01609-2280, USA
Abstract
Dierent surface roughnesses of six chocolate samples were produced by molding over sandpaper of dierent graininess. Surfaces
were examined for roughness (laser scanning microscopy), color and image texture (digital vision system) and gloss (glossmeter). Samples
exhibited signicantly dierent roughness among them expressed by the two parameters used to characterize their surfaces: the statistical
average roughness, ARa (lm), and the area-scale fractal complexity (dimensionless), Asfc. Surfaces of sandpaper and chocolate samples
were highly correlated with these two parameters. Surface elements related to roughness were in the order of 314 lm. Gloss of chocolate
surfaces diminished exponentially as roughness increased while color (L*, lightness and whiteness index) decreased linearly. Parameters
describing image texture, entropy and homogeneity, varied linearly with Asfc values. The structure of the surface of chocolate bars seems
to play a decisive role in visual quality appearance.
2005 Elsevier Ltd. All rights reserved.
0260-8774/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2005.08.004
V. Briones et al. / Journal of Food Engineering 77 (2006) 776783 777
2.4.1.2. Image texture analysis. In order to obtain the tex- 3.1. Changes in roughness parameters
tural features of the image co-occurrence matrices were
generated from the pixel values of gray-level images. The ARa is a common parameter to represent roughness.
co-occurrence matrix (Haralick et al., 1973) is based the Since ARa describes the magnitude of protrusions and
estimation of the second-order joint probability density depressions of the surface (e.g., in lm), it was compared
functions P(i, j, d, h), this is, the probability of going from with Asfc, which is a non-dimensional parameter. As
gray level i to gray level j, when inter-sample spacing is d shown in Table 1, ARa values of chocolate samples were
and direction is h (Gonzalez & Woods, 1992; Haralick, signicantly dierent according to ANOVA and LSD tests
1979). The segmented images were converted to grayscale (p < 0.05). ARa values decreased signicantly from
and analyzed using the average of one direction, h = 0 12.1 lm for CH40 to 3.4 lm for CH400. The same trend
and distance, d = 1. Only two textural features, homogene- is observed to occur when the fractal parameter Asfc is
ity and entropy, were extracted from images for the pur- used. Asfc values ranged from 317.7 for the roughest
pose of this study. For example, if all pixels in an image sample (CH40) to 242.5 for the smoothest one CH(400),
have the same value the homogeneity is one and entropy a dierence of 75.2 units. Coincidentally, values of ARa
is zero. are similar to the size of fat crystals on bloomed chocolate
(Adenier, Chaveron, & Ollivon, 1993).
2.5. Statistical analysis The relation between surface roughness of sandpaper
and that of chocolate samples expressed by the two rough-
The statistical analysis of variance (ANOVA) and test of ness parameters (ARa and Asfc) are shown in Fig. 2. The
least signicant dierences of means (LSD) were conduced correlation coecient was higher for Asfc than for ARa
using Statgraphics Plus for Windows software, Version 5.1 (R2 = 0.99 and 0.89, respectively), a result that suggests
(Manugistic, Inc., Rockville, MD, USA). that the area-scale fractal complexity may be a better
parameter than average roughness to quantify the rough-
3. Results and discussion ness of chocolate samples.
The correlation between the roughness of chocolate
A gallery of digital images of chocolate samples showing samples measured by ARa and by Asfc was quite high
their surface appearance is shown in Fig. 1. The surface (R2 = 0.95) as shown in Fig. 3. From this good correlation
topography of chocolate samples varied appreciably from it may be inferred that the surface features causing the
smooth (CH400, Fig. 1a) to rough (CH40, Fig. 1f), as fractal complexity associated to Asfc are in the size scale
can be appreciated with the naked eye (Fig. 1). of 314 lm (Table 1).
Fig. 1. Images of the rough side of chocolate samples: (a) CH400, (b) CH320, (c) CH220, (d) CH180, (e) CH80 and (f) CH40. Numbers represent
sandpaper coarseness: #40 (coarse), #80 (medium), #180 (ne), #220 (very ne), #320 (extra ne) and #400 (super ne).
780 V. Briones et al. / Journal of Food Engineering 77 (2006) 776783
Table 1
Average surface roughness (ARa) and area-scale fractal complexity (Asfc) of chocolate samples (mean standard deviation)
Parameters Chocolate samples
CH40 CH80 CH180 CH220 CH320 CH400
ARa (lm) 12.1 1.31a 10.0 0.51b 8.0 0.21c 6.9 0.25d 5.7 0.21e 3.4 0.23f
Asfc 317.7 3.88a 298.3 5.22b 283.1 4.44c 270.2 5.80d 258.6 6.54e 242.5 5.11f
af
Means within a row with the same letter are not signicantly dierent (p < 0.05).
16 400
14 375
40 40
12 350
80
(Sandpaper)
80
(Sandpaper)
10 325
ARa ( m)
180
Asfc
8 180 300
220
220
6 275
320 320
4 250
400 400
2 225
R2 = 0.89 R2 = 0.99
0 200
0 2 4 6 8 10 12 14 16 200 225 250 275 300 325 350 375 400
ARa ( m) Asfc
a (Chocolate samples) b (Chocolate samples)
Fig. 2. Relation between surface roughness (expressed as ARa or Asfc) of chocolate samples and sandpaper. Squares and triangles indicate values of
individual measurements on chocolate samples and dotted lines enclose values for each sandpaper type.
Table 2
Average surface gloss (85 angle) of chocolate samples (mean standard deviation)
Parameters Chocolate samples
CH40 CH80 CH180 CH220 CH320 CH400
85 Gloss (GU) 0.87 0.20a 1.40 0.05b 2.01 0.09c 2.71 0.26d 3.22 0.13e 4.98 0.13f
af
Means with dierent letters are signicantly dierent (p < 0.05).
V. Briones et al. / Journal of Food Engineering 77 (2006) 776783 781
6 50 46
CH400
CH400 48 44
CH320
5 2 42
R = 0.96 46 CH220 R2 = 0.95 40
85 Gloss (G.U.)
CH180
CH320 44
4 38
CH400 CH80
CH220 42 36
L*
CH40
WI
3 CH320
CH180 40 34
CH220 32
2 CH80 38 CH180
R2 = 0.95 30
CH40 36 CH80 28
1
34 CH40 26
0 32 24
225 250 275 300 325 350
225 250 275 300 325 350
Asfc
Asfc
Fig. 5. L* and WI versus Asfc. Squares and triangles indicate values of
Fig. 4. Relation between Asfc and gloss (85) in chocolate samples.
individual measurements.
Squares indicate values of individual measurements.
Table 3
Average surface L*, a*, b* color parameters and WI values (and standard deviation) for chocolate samples
Parameters Chocolate samples
CH40 CH80 CH180 CH220 CH320 CH400
L* 35 0.11a 36 0.24b 39 0.13c 40 0.05d 41 0.49e 45 0.16f
a* 6 0.36a 6 0.03b 7 0.27c 7 0.15c 8 0.05d 8 0.09d
b* 7 0.28a 8 0.01b 8 0.10b 9 0.48c 9 0.06c 10 0.06d
WI 34 0.08a 36 0.24b 38 0.14c 39 0.05d 40 0.47e 43 0.14f
af
Means within a row with the same letter are not signicantly dierent (p < 0.05).
782 V. Briones et al. / Journal of Food Engineering 77 (2006) 776783
Table 4
Average surface entropy and homogeneity values and standard deviation depending on the chocolate supercial roughness
Parameters Chocolate samples
CH40 CH80 CH180 CH220 CH320 CH400
Entropy 2.60 0.05a 2.23 0.02b 1.96 0.05c 1.83 0.08d 1.74 0.09e 1.59 0.05f
Homogeneity 0.31 0.014a 0.37 0.003b 0.46 0.012c 0.50 0.006d 0.53 0.006e 0.56 0.006f
af
Means within a row with the same letter are not signicantly dierent (p < 0.05).
3.1
studies of foods and beverages (pp. 353389). Amsterdam, The
0.6
CH320
R2 = 0.97
Netherlands.
CH220
CH400 ASTM (1995). Standard test method for specular gloss. Designation D
CH180
CH40 523. In 1995 Annual book of ASTM standards. Volume 6.01: Paint-tests
CH80
for chemical, physical and optical properties; appearance; durability of
2.5 0.4
Homogeneity
non-metallic materials. Philadelphia: American Society for Testing and
Entropy
CH40 Materials.
CH80 Augusteijn, M. F., Clemens, L. E., & Shaw, K. A. (1995). Performance
R2 = 0.96
CH180
evaluation of texture measures for ground cover identication in
1.9 0.2 satellite images by means of a neural network classier. IEEE
CH220
Transactions on Geoscience and Remote Sensing, 33(3), 616626.
CH320
Barnett, S. (1973). Freezing of coee extract to produce a dark colored
CH400 freeze-dried product. Engineering of food preservation and biochem-
1.3 0 ical processes. AIChE Symposium Series, 69(132), 2632.
225 250 275 300 325 350
Asfc Beckett, S. T. (2002). The science of chocolate (1st ed.). Cambridge: Royal
Society of Chemistry.
Fig. 6. Relation between surface roughness (chocolate) and image texture Boutin, B. (1997). Finishing agents. The Manufacturing Confectioner,
(entropy and homogeneity). Squares and triangle indicate values of 68(10), 102109.
individual measurements. Briones, V., & Aguilera, J. M. (2005). Image analysis of changes in surface
color of chocolate. Food Research International, 38, 8794.
Brown, C. A., Charles, P. D., Johnsen, W. A., & Chestera, S. (1993).
Fractal analysis of topographic data by the Patchwork method. Wear,
(R2 = 0.96 and R2 = 0.97, respectively) and with the visual 161, 6167.
appearance of the images, validates the usefulness of the Brown, C. A., Johnsen, W. A., & Hult, K. M. (1998). Scale sensitivity,
methodology. fractal analysis and simulations. International Journal of Machine
Tools & Manufacture, 38(56), 633637.
4. Conclusions BYK. (1997). Haze-gloss. Application note. Accompanying handout to
color and appearance. In Seminar given by BYK-Garden USA, Foster
City, California.
Laser microscopy was instrumental to generate topo- Dalal, E. N., & Natale-Homan, K. M. (1999). The eect of gloss on
graphical data of the surface of chocolate at the required color. Color Research and Application, 24, 369376.
micrometer scale. The surface roughness of chocolate sam- Eliades, T., Gioka, C., Heim, M., Eliades, G., & Makou, M. (2004). Color
ples having the same composition but imprinted with sand- stability of orthodontic adhesive resins. The Angle Orthodontist, 74(3),
391393.
paper of dierent graininess had a major eect on the Gonzalez, R. C., & Woods, R. E. (1992). Digital image processing (pp.
measured gloss and color. Good correlations were found 282325) (second ed.). Boston, MA: Addison-Wesley Longman
between roughness parameters (Asfc and ARa) and gloss Publishing Co. Inc..
value, lightness (L*), whiteness index and image texture Haralick, R. M., Shanmugan, K., & Dinstein, I. (1973). Textural features
features such as homogeneity and entropy. Further work for image classication. IEEE Transactions on Systems, Man, and
Cybernetics, SMC-3(6), 610621.
is needed to relate surface roughness of commercially tem- Haralick, R. M. (1979). Statistical and structural approach to texture.
pered and molded chocolate and these key quality Proceeding of the IEEE, 67(5), 786804.
parameters. Hatcher, D. W., Symons, S. J., & Manivannan, U. (2004). Developments
in the use of image analysis for the assessment of oriental noodle
Acknowledgements appearance and colour. Journal of Food Engineering, 61, 109117.
Jahns, G., Nielsen, H. M., & Paul, W. (2001). Measuring image analysis
attributes and modelling fuzzy consumer aspects for tomato quality
This work was partly funded by the Nestle Research granding. Computers and Electronics in Agricultura, 31, 1729.
Centre (Vers-chez-les-Blanc, Switzerland) and a graduate Keyf, F., & Etikan, I. (2004). Evaluation of gloss changes of two denture
fellowship to Ms. V. Briones by the Oce of Scientic acrylic resin materials in tour dierent beverages. Dental Materials,
Research (DIPUC), Universidad Catolica de Chile. 20(3), 244251.
Kim, I., Lee, Y., Lim, B., & Kim, C. (2003). Eect of surface topography
on the color of dental porcelain. Journal of Materials Science:
References Materials in Medicine, 14, 405409.
Lee, S. Y., Dangaran, K. L., & Krochta, J. M. (2002). Gloss stability of
Adenier, H., Chaveron, H., & Ollivon, M. (1993). Mechanism of fat whey-protein/plasticizer coating formulations on chocolate surface.
bloom development on chocolate. In G. Charalambous (Ed.), Shelf life Journal of Food Science, 67, 11211125.
V. Briones et al. / Journal of Food Engineering 77 (2006) 776783 783
Leemans, V., Magein, H., & Destain, M. (1998). Defects segmentation on Pickett, R. M. (1970). Visual analysis of texture in the detection and
Golden Delicious apples by using colour machine vision. Computer recognition of objects. In B. Lipkin, & A. Rosenfeld (Eds.), Picture
and Electronics in Agriculture, 20, 117130. processing & psychopictorics (pp. 298308). Orlando.
Lohman, M., & Hartel, R. (1994). Eect of milk fat fractions on fat in Simonot, L., & Elias, M. (2002). Color changes due to surface state
dark chocolate. Journal American Oil Chemists Society, 71(3), modication. Color Research and Application, 28(1), 4549.
267275. Tan, F. J., Morgan, M. T., Ludas, L. I., Forrest, J. C., & Gerrard, D. C.
McGunnigle, G., & Chantler, M. (1999). Rotation invariant classication (2000). Assessment of fresh por colour with colour machine vision.
or rough surfaces. IEEE Proceedings Vision Image & Signal Processing, Journal of Animal Science, 78(12), 30783085.
146, 345352. Thomas, T. (1999). Rough surfaces (2nd ed.). London: Imperial College
Pedreschi, F., Aguilera, J. M., & Brown, C. (2000). Characterization of Press.
food surfaces using scale-sensitive fractal analysis. Journal of Food Trezza, T. A., & Krochta, J. M. (2000). The gloss of edible coatings as
Process Engineering, 23, 127143. aected by surfactants, lipids, relative humidity and time. Journal of
Pedreschi, F., Aguilera, J. M., & Brown, C. (2002). Characterization Food Science, 65, 658662.
of the surface properties of chocolates using scale-sensitive Yam, K. L., & Papakadis, S. E. (2004). A simple digital imaging method
fractal analysis. International Journal of Food Properties, 5(3), for measuring and analysing color of food surfaces. Journal of Food
523535. Engineering, 61(1), 137142.