- Email: [email protected]
Monitoring pasta production line using automated imaging technique
Available online at www.sciencedirect.com Available online at www.sciencedirect.com
Procedia Food Science
Procedia – Food (2011)1173 000–000 Procedia FoodScience Science 100(2011) – 1180 www.elsevier.com/locate/procedia
11th International Congress on Engineering and Food (ICEF11)
Monitoring pasta production line using automated imaging technique A.Mokhtar*a, M.A.Hussein, T. Becker Group of (Bio-) Process Technology and Process Analysis, Faculty of Life Science Engineering, Technische Universität München, Weihenstephaner Steig 20, 85354 Freising, Germany.
Abstract In food mass production visual inspection does not match the high standards exhibited in food technology, not to mention the economic costs and drawback from the human factor involved which mainly burdens quality assurance. In focus, quality control, on a daily basis, is just performed on selected samples which are thought to indicate the whole production trend. Therefore an automated solution should be used, which offers an inline monitoring protocol joined with more statistical measurement evaluation. Pasta is considered one of the typical food around the world, production quality of pasta have always been of concern to producers. Pasta, in order to meet the criteria and expectations of consumers, must avoid the production defects like crumbling during packaging, in-homogeneousness, unevenness in size, cracks due to extra dehydration, additionally must not be glutinous on the surface (stick together). Process analytical technology is a time saver, precise and economic solution which the industry can rely on. In this work a pasta production line monitoring is developed conceding to PAT basis (process analytical technology); using an industrial camera controlled by microcontroller, and a developed combinatorial algorithm using image processing techniques. Glutinous is easily detected by intelligent edge detection algorithm and pasta in-homogeneousness defect is evaluated by automated surface roughness evaluation. © Published by Elsevier SelectionLtd. and/orSelection peer-review and/or under responsibility of 11thunder International Congress ©2011 2011 Published byB.V. Elsevier peer-review responsibility on Engineering and Food (ICEF 11) Executive Committee. Executive Committee Members
of ICEF11
Keywords: Edge Detection; Surface Roughness; Microcontroller; Pasta Production Defects.
1. Introduction In some food industry, quality evaluation is performed manually by visual inspection, which is tedious, laborious, costly and inherently unreliable due to its subjective nature [1]. The increase in awareness and consumers requirements have created the expectation or improved quality in consumer food products [2]. Consequently Francis (1980) found that human perception could be easily fooled. With the high labour
* Corresponding author. Tel.: +49-8161-71-2626; fax: +49-8161-71-3883. E-mail address: [email protected] .
2211–601X © 2011 Published by Elsevier B.V. Selection and/or peer-review under responsibility of 11th International Congress on Engineering and Food (ICEF 11) Executive Committee. doi:10.1016/j.profoo.2011.09.175
1174
A.Mokhtar al. / Procedia Food Science (2011)000–000 1173 – 1180 A.Mokhtar/etProcedia – Food Science 00 1(2011)
costs, inconsistency and variability associated with human inspection emphasizes the need for objective measurements systems [3]. The food industry considered one of the top ten industries using image processing techniques [4]. Which is non-invasive techniques for evaluation the food products. Recently, computer vision employing image processing techniques has been developed rapidly, which can quantitatively characterize complex geometrical and textural features of foods. Image processing algorithms play an important role in the food quality evaluation by maintaining accuracy and consistency while eliminating the subjectivity of visual inspections [1]. Advances in hardware and software have aided in the expansion of machine vision in food quality inspection by providing low cost powerful solutions, leading to more studies on the development of computer vision systems in the food industry [5, 6]. Computer vision is the features of clear and expressive descriptions of physical objects from images [7]. A computer vision system generally consists of five basic components: illumination, a camera, an image capture software, computer [8]. In most cases manual inspections are time-consuming; moreover the accuracy of the inspection cannot be guaranteed [9]. By contrast it has been found that computer vision inspection of food products was more consistent, efficient and cost effective [10, 11]. In this work several pattern recognition techniques were developed in order to classify pasta defects, such as stickiness, unevenness in size, broken pieces (crumbles) and cracks. In this work pattern recognition and crack detection techniques were developed in order to classify pasta defects, such as stickiness, unevenness in size, broken pieces (crumbles) and cracks. 2. Materials & Methods 2.1. Pasta samples In this work algorithms are applied on two types of pasta, Spätzle (ribbon-cut egg noodles) which has several defects as listed in Table1.; and Bucatini (Hollow-spaghetti) which suffers from cracks due to unbalanced drying conditions. In the experiments, statistical study is done on 177 stickiness samples, 261 crumbles samples and 478 unevenness in size samples. Images are taken for defected samples, and then converted to binary. In this study all objects features extracted from the binary image are in pixels. 2.2. Imaging setup Flat LED light supported by a white diffusion filter, Basler piA2400-17gm/gc CCD camera 5MP, 17fps which is satisfactory in still images. Preliminary setup was used for validation and test, for further work microcontroller will be used as shown in Figure 1 to enable online processing. Pasta defects are classified into several categories from a specified by pasta manufacturer in Germany as listed in Table 1.
Fig. 1. Pasta defects (a) Cracks, (b) crumbles, (c) Unevenness in size and (d) Stickiness
1175
A.Mokhtar al. / Procedia Food Science 1 (2011) 1173 000–000 – 1180 Authoretname / Procedia – Food Science 00 (2011) Table 1. Pasta defects Defect
Reason
Image
Cracks
Unbalanced drying conditions
Fig 1 (a)
Broken pieces and crumbles
Mechanical stress
Fig 1 (b)
Unevenness in size
Variation of dough-flow during extrusion
Fig 1 (c)
Stickiness (sticky surface)
Withdrawing starch during cooking
Fig 1 (d)
2.3. Object measurement Measurements that can be performed on features in images for food quality evaluation can be classified into four categories: size, shape, color, and texture [1]. In this study area, perimeter and diameter represent the size category; aspect ratio represents the shape, and the RGB represent the color. 2.3.1. Area Area is measured from the binary image in pixels, by counting the number of pixels inside the object; in further study area will be calculated in meter. 2.3.2. Aspect ratio Aspect ratio itself cannot differentiate among different sizes, therefore in this work it combined with area and perimeter to present a meaningful result. Aspect Ratio
MaxDiameter MinDiameter
(1)
Fig. 2. Error handling: (a) stickiness sample, (b) digital representation (c) error source in focus and (d) digital representation of error source in focus
2.3.3. Perimeter Using lattice see Fig. 3 method then propagation takes place to search for the next point on the surface. Some problems were faced in application of algorithm on the binary image, such as sticking in a closed loop inside the pasta tiny cavities as shown in Fig. 2; such problems are solved by tagging the counted pixel in the perimeter to avoid passing through the same pixel twice.
1176
A.Mokhtar et al. / Procedia Food Science (2011)000–000 1173 – 1180 A.Mokhtar/ Procedia – Food Science 001(2011)
Fig. 3. Lattice
2.3.4. Compactness and Roundness Using compactness and roundness is not useful in defect classification, as shown in Fig. 5d, e
Compactness
Roundness
4 . Area Perimeter 2
4 . Area . MaxDiameter 2
(2)
(3)
2.3.5. Surface roughness The conventional approach for surface roughness estimation based up on either the first derivative (gradient) (4) of the image or evaluating the second derivative (Laplacian) (5) [13], both are discretized second order in space (6) and (7).
2
df fi 1 fi 1 2 x dx
(4)
d 2 f fi 1 2 fi f i 1 x 2 dx 2
(5)
f i 1 f i 1 f j 1 f j 1 2 x 2 y
(6)
f 2 f j f j 1 fi 1 2 fi fi 1 j 1 2 x y 2
(7)
2.4. Edge detection For the capturing of the pasta significance area an edge detection algorithm based upon Skewness and energy patterns [12] and implemented in an in-house developed software ImgPro V 0.4, after the significance area is predicted several parameter are evaluated for pasta faulty predictions.
1177
A.Mokhtar et name al. / Procedia Food Science 1 (2011) 1173 –000–000 1180 Author / Procedia – Food Science 00 (2011)
Fig. 4. Flow chart: (a) determination of the classifying thresholds and (b) using threshold to classify defects
Flow charts of prediction algorithm and combination of features in order to point out the pasta faults is shown in fig. 4. 3. Results & Discussion Using different geometrical values, such as area, perimeter, aspect ratio, compactness and roundness to classify the pasta defects listed in Table1, Area is a good feature used to determine the stickiness. Perimeter differentiates between crumbles and unevenness in size defects. Roundness and compactness are inadequate to identify a specific defect as shown in fig. 5(d, e) because the values overlap.
Perimeter (in pixel)
10000
1000
100
10
0
50
100
Sample index
150
200
1178
A.Mokhtar et al. / Procedia FoodScience Science001 (2011) 000–000 1173 – 1180 A.Mokhtar/ Procedia – Food
Fig. 5. Logarithmic representation for (a) area , (b) perimeter, (c) aspect ratio, (d) compactness and (e) Roundness Vs several defected samples
Surface roughness algorithm is used to determine the cracks in pasta. fig. 6 shows the first derivative of two samples. Napla R , Napla G and Napla B is the red, green, blue component of color intensity gradient.
Fig. 6. Surface roughness components (R,G and B) for two cracked samples
A.Mokhtar et al. / Procedia Food– Science 1 (2011) 1180 Author name / Procedia Food Science 00 1173 (2011)– 000–000
Perimeter threshold at 250
Area threshold at 800
Fig. 7. 3D plot in area, perimeter, Thresholds classifying stickiness, unevenness and crumbles
3D representation for the samples combined using support vector machine to specifically classify pasta defects as shown in fig. 7. Area, perimeter and aspect ratio form 3D axes of the graph. Two threshold planes determine can distinguish among three types of pasta defects stickiness, unevenness in size and crumbles with an error of 1% for stickiness 5% crumbles and 13% for unevenness. 4. Conclusions In industry using visual inspection is not quite efficient, therefore the need for an automated inspection supported by an intelligent algorithms are obligatory to fulfill the customers quality required. Pasta defects analyzed in this study are stickiness, cracks, crumbles and unevenness in size. Applying support vector machine to 3D graph in area, perimeter and aspect ratio; pasta defects were classified. The error in determining the stickiness defect is 1%, crumples error is 5% and unevenness in size error is 13%. Crack defect is recognized by the gradient in color intensity value. Future work is to calculate object characteristics with more amounts of samples to obtain good statistical results, online video processing controlled by a microcontroller to be useful for industry, also to enhance algorithms in order to reduce errors.
1179
1180
A.Mokhtar et al. / Procedia FoodScience Science00 1 (2011) 1173 – 1180 A.Mokhtar/ Procedia – Food 000–000
References [1] Du C.-J., Sun D.W. 2003. Recent developments in the applications of image processing techniques for food quality evaluation. Journal of Food Science and Technology, 15(2004), 230-249. [2] Brosnan T., Sun D.W. 2003. Improving quality inspection of food products by computer vision––a review. Journal of Food Engineering 61 ; 2004, 3–16. [3] Francis, F. J. Colour quality evaluation of horticultural crops. HortScience; 1980, 15(1), 14–15. [4] Gunasekaran, S. Computer vision technology for food quality assurance. Trends in Food Science Technology; 1996, 7, 245– 256. [5] Locht, P., Thomsen, K., & Mikkelsen, P. Full color image analysis as a tool for quality control and process evelopment in the food industry. In 1997 ASAE Annual International Meeting, Paper No. 973006. St. Joseph, Michigan, USA: ASAE. [6] Sun, D.-W. Inspecting pizza topping percentage and distribution by a computer vision method. Journal of Food Engineering; 2000, 44, 245–249. [7] Ballard, D. A., Brown, C. M. Computer vision. Englewood Cliffs, NJ, USA: Prentice-Hall; 1982. [8] Wang, H.-H., Sun, D.-W. Correlation between cheese meltability determined with a computer vision method and with Arnott and Schreiber. Journal of Food Science; 2002a, 67(2), 745–749. [9] Park, B., Chen, Y.R., Nguyen, M., Hwang, H. Characterising multispectral images of tumorous, bruised, skin-torn, znd wholesome poultry carcasses. Transactions of the ASAE, 39(5), pp. 1933-1941, 1996. [10].Lu, J., Tan, J., Shatadal, P., Gerrard, D.E. Evaluation of pork colour by using computer vision. Meat Science 56; 2000, pp. 57-60. [11] Tao, Y., Heinemann, P.H., Varghese, Z., Morrow, C.T., Sommer, H.J. Machine vision for colour inspection of potatoes and apples. Transactions of the SAE, 38(5), pp. 1555-1561, 1995a. [12] Hussein, W. B., Moaty, A. A., Hussein, M. A., Becker, T. A Novel Edge Detection Method Based On Image Energy And Skewness With Application To Intramuscular Fat Recognition. Proceeding of IADIS Multi-Conference on Computer Science and Information Systems 2010; 27-29 July 2010. [13] Hussein, M.A. and Becker, T. Numerical modelling of shear and normal stress of micro-porous ceramics for stimulated invitro cultivation of bone cells. Microfluid Nanofluid. 2009, DOI 10.1007/s10404-009-0499-1.
Presented at ICEF11 (May 22-26, 2011 – Athens, Greece) as paper MCF735.
Procedia Food Science
Procedia – Food (2011)1173 000–000 Procedia FoodScience Science 100(2011) – 1180 www.elsevier.com/locate/procedia
11th International Congress on Engineering and Food (ICEF11)
Monitoring pasta production line using automated imaging technique A.Mokhtar*a, M.A.Hussein, T. Becker Group of (Bio-) Process Technology and Process Analysis, Faculty of Life Science Engineering, Technische Universität München, Weihenstephaner Steig 20, 85354 Freising, Germany.
Abstract In food mass production visual inspection does not match the high standards exhibited in food technology, not to mention the economic costs and drawback from the human factor involved which mainly burdens quality assurance. In focus, quality control, on a daily basis, is just performed on selected samples which are thought to indicate the whole production trend. Therefore an automated solution should be used, which offers an inline monitoring protocol joined with more statistical measurement evaluation. Pasta is considered one of the typical food around the world, production quality of pasta have always been of concern to producers. Pasta, in order to meet the criteria and expectations of consumers, must avoid the production defects like crumbling during packaging, in-homogeneousness, unevenness in size, cracks due to extra dehydration, additionally must not be glutinous on the surface (stick together). Process analytical technology is a time saver, precise and economic solution which the industry can rely on. In this work a pasta production line monitoring is developed conceding to PAT basis (process analytical technology); using an industrial camera controlled by microcontroller, and a developed combinatorial algorithm using image processing techniques. Glutinous is easily detected by intelligent edge detection algorithm and pasta in-homogeneousness defect is evaluated by automated surface roughness evaluation. © Published by Elsevier SelectionLtd. and/orSelection peer-review and/or under responsibility of 11thunder International Congress ©2011 2011 Published byB.V. Elsevier peer-review responsibility on Engineering and Food (ICEF 11) Executive Committee. Executive Committee Members
of ICEF11
Keywords: Edge Detection; Surface Roughness; Microcontroller; Pasta Production Defects.
1. Introduction In some food industry, quality evaluation is performed manually by visual inspection, which is tedious, laborious, costly and inherently unreliable due to its subjective nature [1]. The increase in awareness and consumers requirements have created the expectation or improved quality in consumer food products [2]. Consequently Francis (1980) found that human perception could be easily fooled. With the high labour
* Corresponding author. Tel.: +49-8161-71-2626; fax: +49-8161-71-3883. E-mail address: [email protected] .
2211–601X © 2011 Published by Elsevier B.V. Selection and/or peer-review under responsibility of 11th International Congress on Engineering and Food (ICEF 11) Executive Committee. doi:10.1016/j.profoo.2011.09.175
1174
A.Mokhtar al. / Procedia Food Science (2011)000–000 1173 – 1180 A.Mokhtar/etProcedia – Food Science 00 1(2011)
costs, inconsistency and variability associated with human inspection emphasizes the need for objective measurements systems [3]. The food industry considered one of the top ten industries using image processing techniques [4]. Which is non-invasive techniques for evaluation the food products. Recently, computer vision employing image processing techniques has been developed rapidly, which can quantitatively characterize complex geometrical and textural features of foods. Image processing algorithms play an important role in the food quality evaluation by maintaining accuracy and consistency while eliminating the subjectivity of visual inspections [1]. Advances in hardware and software have aided in the expansion of machine vision in food quality inspection by providing low cost powerful solutions, leading to more studies on the development of computer vision systems in the food industry [5, 6]. Computer vision is the features of clear and expressive descriptions of physical objects from images [7]. A computer vision system generally consists of five basic components: illumination, a camera, an image capture software, computer [8]. In most cases manual inspections are time-consuming; moreover the accuracy of the inspection cannot be guaranteed [9]. By contrast it has been found that computer vision inspection of food products was more consistent, efficient and cost effective [10, 11]. In this work several pattern recognition techniques were developed in order to classify pasta defects, such as stickiness, unevenness in size, broken pieces (crumbles) and cracks. In this work pattern recognition and crack detection techniques were developed in order to classify pasta defects, such as stickiness, unevenness in size, broken pieces (crumbles) and cracks. 2. Materials & Methods 2.1. Pasta samples In this work algorithms are applied on two types of pasta, Spätzle (ribbon-cut egg noodles) which has several defects as listed in Table1.; and Bucatini (Hollow-spaghetti) which suffers from cracks due to unbalanced drying conditions. In the experiments, statistical study is done on 177 stickiness samples, 261 crumbles samples and 478 unevenness in size samples. Images are taken for defected samples, and then converted to binary. In this study all objects features extracted from the binary image are in pixels. 2.2. Imaging setup Flat LED light supported by a white diffusion filter, Basler piA2400-17gm/gc CCD camera 5MP, 17fps which is satisfactory in still images. Preliminary setup was used for validation and test, for further work microcontroller will be used as shown in Figure 1 to enable online processing. Pasta defects are classified into several categories from a specified by pasta manufacturer in Germany as listed in Table 1.
Fig. 1. Pasta defects (a) Cracks, (b) crumbles, (c) Unevenness in size and (d) Stickiness
1175
A.Mokhtar al. / Procedia Food Science 1 (2011) 1173 000–000 – 1180 Authoretname / Procedia – Food Science 00 (2011) Table 1. Pasta defects Defect
Reason
Image
Cracks
Unbalanced drying conditions
Fig 1 (a)
Broken pieces and crumbles
Mechanical stress
Fig 1 (b)
Unevenness in size
Variation of dough-flow during extrusion
Fig 1 (c)
Stickiness (sticky surface)
Withdrawing starch during cooking
Fig 1 (d)
2.3. Object measurement Measurements that can be performed on features in images for food quality evaluation can be classified into four categories: size, shape, color, and texture [1]. In this study area, perimeter and diameter represent the size category; aspect ratio represents the shape, and the RGB represent the color. 2.3.1. Area Area is measured from the binary image in pixels, by counting the number of pixels inside the object; in further study area will be calculated in meter. 2.3.2. Aspect ratio Aspect ratio itself cannot differentiate among different sizes, therefore in this work it combined with area and perimeter to present a meaningful result. Aspect Ratio
MaxDiameter MinDiameter
(1)
Fig. 2. Error handling: (a) stickiness sample, (b) digital representation (c) error source in focus and (d) digital representation of error source in focus
2.3.3. Perimeter Using lattice see Fig. 3 method then propagation takes place to search for the next point on the surface. Some problems were faced in application of algorithm on the binary image, such as sticking in a closed loop inside the pasta tiny cavities as shown in Fig. 2; such problems are solved by tagging the counted pixel in the perimeter to avoid passing through the same pixel twice.
1176
A.Mokhtar et al. / Procedia Food Science (2011)000–000 1173 – 1180 A.Mokhtar/ Procedia – Food Science 001(2011)
Fig. 3. Lattice
2.3.4. Compactness and Roundness Using compactness and roundness is not useful in defect classification, as shown in Fig. 5d, e
Compactness
Roundness
4 . Area Perimeter 2
4 . Area . MaxDiameter 2
(2)
(3)
2.3.5. Surface roughness The conventional approach for surface roughness estimation based up on either the first derivative (gradient) (4) of the image or evaluating the second derivative (Laplacian) (5) [13], both are discretized second order in space (6) and (7).
2
df fi 1 fi 1 2 x dx
(4)
d 2 f fi 1 2 fi f i 1 x 2 dx 2
(5)
f i 1 f i 1 f j 1 f j 1 2 x 2 y
(6)
f 2 f j f j 1 fi 1 2 fi fi 1 j 1 2 x y 2
(7)
2.4. Edge detection For the capturing of the pasta significance area an edge detection algorithm based upon Skewness and energy patterns [12] and implemented in an in-house developed software ImgPro V 0.4, after the significance area is predicted several parameter are evaluated for pasta faulty predictions.
1177
A.Mokhtar et name al. / Procedia Food Science 1 (2011) 1173 –000–000 1180 Author / Procedia – Food Science 00 (2011)
Fig. 4. Flow chart: (a) determination of the classifying thresholds and (b) using threshold to classify defects
Flow charts of prediction algorithm and combination of features in order to point out the pasta faults is shown in fig. 4. 3. Results & Discussion Using different geometrical values, such as area, perimeter, aspect ratio, compactness and roundness to classify the pasta defects listed in Table1, Area is a good feature used to determine the stickiness. Perimeter differentiates between crumbles and unevenness in size defects. Roundness and compactness are inadequate to identify a specific defect as shown in fig. 5(d, e) because the values overlap.
Perimeter (in pixel)
10000
1000
100
10
0
50
100
Sample index
150
200
1178
A.Mokhtar et al. / Procedia FoodScience Science001 (2011) 000–000 1173 – 1180 A.Mokhtar/ Procedia – Food
Fig. 5. Logarithmic representation for (a) area , (b) perimeter, (c) aspect ratio, (d) compactness and (e) Roundness Vs several defected samples
Surface roughness algorithm is used to determine the cracks in pasta. fig. 6 shows the first derivative of two samples. Napla R , Napla G and Napla B is the red, green, blue component of color intensity gradient.
Fig. 6. Surface roughness components (R,G and B) for two cracked samples
A.Mokhtar et al. / Procedia Food– Science 1 (2011) 1180 Author name / Procedia Food Science 00 1173 (2011)– 000–000
Perimeter threshold at 250
Area threshold at 800
Fig. 7. 3D plot in area, perimeter, Thresholds classifying stickiness, unevenness and crumbles
3D representation for the samples combined using support vector machine to specifically classify pasta defects as shown in fig. 7. Area, perimeter and aspect ratio form 3D axes of the graph. Two threshold planes determine can distinguish among three types of pasta defects stickiness, unevenness in size and crumbles with an error of 1% for stickiness 5% crumbles and 13% for unevenness. 4. Conclusions In industry using visual inspection is not quite efficient, therefore the need for an automated inspection supported by an intelligent algorithms are obligatory to fulfill the customers quality required. Pasta defects analyzed in this study are stickiness, cracks, crumbles and unevenness in size. Applying support vector machine to 3D graph in area, perimeter and aspect ratio; pasta defects were classified. The error in determining the stickiness defect is 1%, crumples error is 5% and unevenness in size error is 13%. Crack defect is recognized by the gradient in color intensity value. Future work is to calculate object characteristics with more amounts of samples to obtain good statistical results, online video processing controlled by a microcontroller to be useful for industry, also to enhance algorithms in order to reduce errors.
1179
1180
A.Mokhtar et al. / Procedia FoodScience Science00 1 (2011) 1173 – 1180 A.Mokhtar/ Procedia – Food 000–000
References [1] Du C.-J., Sun D.W. 2003. Recent developments in the applications of image processing techniques for food quality evaluation. Journal of Food Science and Technology, 15(2004), 230-249. [2] Brosnan T., Sun D.W. 2003. Improving quality inspection of food products by computer vision––a review. Journal of Food Engineering 61 ; 2004, 3–16. [3] Francis, F. J. Colour quality evaluation of horticultural crops. HortScience; 1980, 15(1), 14–15. [4] Gunasekaran, S. Computer vision technology for food quality assurance. Trends in Food Science Technology; 1996, 7, 245– 256. [5] Locht, P., Thomsen, K., & Mikkelsen, P. Full color image analysis as a tool for quality control and process evelopment in the food industry. In 1997 ASAE Annual International Meeting, Paper No. 973006. St. Joseph, Michigan, USA: ASAE. [6] Sun, D.-W. Inspecting pizza topping percentage and distribution by a computer vision method. Journal of Food Engineering; 2000, 44, 245–249. [7] Ballard, D. A., Brown, C. M. Computer vision. Englewood Cliffs, NJ, USA: Prentice-Hall; 1982. [8] Wang, H.-H., Sun, D.-W. Correlation between cheese meltability determined with a computer vision method and with Arnott and Schreiber. Journal of Food Science; 2002a, 67(2), 745–749. [9] Park, B., Chen, Y.R., Nguyen, M., Hwang, H. Characterising multispectral images of tumorous, bruised, skin-torn, znd wholesome poultry carcasses. Transactions of the ASAE, 39(5), pp. 1933-1941, 1996. [10].Lu, J., Tan, J., Shatadal, P., Gerrard, D.E. Evaluation of pork colour by using computer vision. Meat Science 56; 2000, pp. 57-60. [11] Tao, Y., Heinemann, P.H., Varghese, Z., Morrow, C.T., Sommer, H.J. Machine vision for colour inspection of potatoes and apples. Transactions of the SAE, 38(5), pp. 1555-1561, 1995a. [12] Hussein, W. B., Moaty, A. A., Hussein, M. A., Becker, T. A Novel Edge Detection Method Based On Image Energy And Skewness With Application To Intramuscular Fat Recognition. Proceeding of IADIS Multi-Conference on Computer Science and Information Systems 2010; 27-29 July 2010. [13] Hussein, M.A. and Becker, T. Numerical modelling of shear and normal stress of micro-porous ceramics for stimulated invitro cultivation of bone cells. Microfluid Nanofluid. 2009, DOI 10.1007/s10404-009-0499-1.
Presented at ICEF11 (May 22-26, 2011 – Athens, Greece) as paper MCF735.