unified approach in food quality evaluation using ... - Semantic Scholar
UNIFIED APPROACH IN FOOD QUALITY EVALUATION USING MACHINE VISION Rohit R. Parmar1, Kavindra R. Jain2, Dr. Chintan K. Modi3 1 Research Scholar,2Lecturer,3Assistant Professor EC Department,G.H. Patel College of Engineering & Technology. Vallabh Vidyanagar, India 1 [email protected],[email protected],[email protected] 1,2,3
Abstract— The paper presents a unified approach for quality evaluation of food using image processing and machine vision. In this paper basic tool is combination of computer and machine vision for image analysis and processing through which fast and accurate quality is achieved that too with the help of non-destructive method. Machine vision in food has broadened its range of applications from grains, cereals, fruits to vegetables including processed products as well as spices in which there is a high degree of quality achieved as compared to human vision inspection. In this paper we quantify the qualities of various food products and figure out features which are directly or inversely affect the quality of the food product. Based on these features a generalized formula of quality is proposed to be used for quality evaluation of any type of food product. Keywords: Quality, Machine vision, Image processing, Fruit, Vegetables, Grains, Spices, Unification approach.
1.
INTRODUCTION
The agricultural industry is probably too oldest and most widespread industry in the world. Quality control is of major importance in the food industry. After harvesting a food product based on quality parameter a food product has been sorted and graded in different grades. Also a continuous development in mechanical harvesting system and need for automatic grading system has arisen. In recent years due to rapid industrialization and massive rural-urban migration the necessity of agricultural mechanization needs to be grown faster. Non-destructive quality evaluation of food products is an important and very vital factor in food/agricultural industry. Various parameters which define quality of these products (e.g. size, shape, color, texture, external defects, smell etc.) are evaluated by human inspectors [35]. Together with the high labor costs, inconsistency and variability associated with human inspection accentuates the need for objective measurements systems. Efforts are being geared towards the replacement of human operator with automated systems, as human operations are inconsistent and less efficient [14]. Rapid advances in hardware and software for digital image processing have motivated several studies on the development of computer vision systems to evaluate quality of diverse raw and processed foods. Advancement in computer technology leads to use these in the domain of food processing like grading, sorting and quality inspection [32]. Computer vision systems are recognized as the integration of devices for non-contact optical sensing, and computing and decision processes which receive and interpret automatically an image of a real scene. Recently many different features like size, shape, color and texture of food products are combined together for their applications in the food industry. Normally, by increasing the features used, the performance of the methods proposed can be increased. However, in literature there is a need of a common evaluation scheme. In this paper we propose a unification approach to evaluate the quality of food products. In this paper we present a unified approach to evaluate the quality of food products using machine vision. Section II discusses basic machine vision technique for quality evaluation of food. Section III discusses background of various food products in which machine vision is used. We propose a unified approach for evaluation of quality of food products in section IV. Section V concludes the paper. 2.
QUALITY EVALUATION USING MACHINE VISION
Computer & Machine vision system for quality evaluation of food products contains standard hardware configuration as shown in figure 2.1. It consists of : 1. Some means of presenting the object to be inspected to the camera
2. 3. 4. 5.
Light source for proper illumination CCD Camera to acquire image A frame-grabber, to perform analog-to-digital conversion. Software, if a computer is used for image processing
Figure 2.1 Imaging setup of computer vision system [5]
Illumination plays a vital role in quality of image. So here proper illumination is adjusted with the help of available light sources. Image acquisition is done (or image formation) to produce digital images of the food products. Image is captured by high quality CCD camera. Pre-processing is used to enhance, clean, and improve key parts of the input images [15], [16]. During the process of image acquisition, captured images may suffer from severe noise and external disturbances this can be removed with the help of filtering process. Image can be converted from color to gray scale depending upon whether color parameter is of importance or not. Segmentation is done to identify relevant parts of a food product, as disjoint and non-overlapping regions separated from the background. To partition an image, thresholding is applied on image. Object recognition and measurement are done to quantify key features of the selected objects. Here different parameters based on size, shape etc. is calculated. Edge operators detect discontinuities in grey level, color, texture, etc. Region segmentation involves the grouping together of similar pixels to form regions representing single objects within the image. In interpretation, where the features extracted from the products are interpreted using application domain knowledge. 3
ASSESMENT OF FOOD PRODUCTS
Food product images captured by vision system can be used to identify, analyze and quality assessment of food products. (1) Peanuts Peanut is a mass consumption item and is used for extraction of oil, for making butter, chikkies and chocolates, as an ingredient in making several food and various snacks. India is one of the largest producers along with the USA; China and Argentina. Gujarat, Andhra Pradesh, Tamilnadu and Maharashtra are the main cultivating states. Peanuts from the Saurashtra region of Gujarat are famous all over the world on account of their big size, nutty flavor and crunchy taste. Peanuts having a wide range of seed size and maturity are often obtained at harvest. Due to nonuniform maturity levels size variation has been seen in peanuts like a smaller sized seed would be immature seed and a big seed would be mature as shown in figure 3.1 [3] [11].
Figure 3.1 Partial pictures of peanut in different varieties [11]
In 1978, J. I. Davidson derived a mathematical relationship for seed size distribution to describe relation between different seed (arachies hypogaea L.) varieties. A step further in 1989, F. E. Dowell, who developed an automatic peanut grading system based on machine vision system with the integration of mechanical component to grade peanuts based on size or defect [11]. In 2009, H. Zhong-zhi and Z. You-gang compared 48 different peanut varieties. Using principal component analysis they developed an artificial neural network to establish a seed recognition model which was made up of 49 distinct appearance characteristics referred as shape, texture, color etc.. Here they got variety recognition rate and quality recognition rate up to 91.2% and 93.0% respectively [10]. In 2010, H. Zhong-zhi et al developed a neural network, which was based on 52 appearance features. They got success up to 95.6% in their results. This method has been used for appraisal of peanut quality in China. Seeing to the above studies on peanut the conclusion can be drawn that there are few quality parameters affecting the production of peanuts. For instance color (x1), texture (x2) are directly affecting the quality & non uniform maturity level i.e. size (major (y1) and minor (y2) axis) are inversely proportional to quality of peanuts. (2) Rice Rice is one of the leading food crops of the world and more than half of the world’s population relies on rice as the major daily source of calories and protein [4]. The quality of rice has distinct effect, so the proper inspection of rice quality is very important. There are two main factors for checking the quality of rice kernel: percentage of broken rice and percentage of the purity of rice as shown in figure 3.2.
Figure 3.2 Picture of rice characteristics [5]
In 2002, Y. N. Wan developed an automatic grain kernel handling system in which 1296 singularized kernel images per minute were taken for machine vision inspection. He developed a Windows-based software program for rice quality inspection. [33]. Later on S, Sansomboonsuk and N. Afzulpurkar (2008) developed image processing algorithms and used to extract features for kernels [30]. They measured and calculated area, perimeter, circularity and shape compactness as criteria in Fuzzy logic for classifying each kernel. From testing the image analysis algorithms, the results are obtained, they found accuracy averaging 92% for both of the broken rice and the purity of rice compared with human inspection. In 2009, B. Emadzadeh, S.M.A. Razavi, and R. Farahmandfar [9] compared three Iranian rice varieties, namely Tarom Mahalli, Fajr and Neda, determined parameters using micrometer and image processing methods. Comparison of the results obtained by both procedures showed that the geometric characteristics (length, width, height and projected area) of all three varieties decrease and sphericity increases after removing the outer and the brownish layers. It was found that the values of micrometer data are having lower for all the geometric factors and that the true size and sphericity. In 2010, S. Shantaiya, U. Ansari developed digital image analysis algorithm based on color, morphological and textural features to identify the six varieties rice seeds in Chhattisgarh region [29]. Nine color and nine morphological and textural features were used for analysis. They developed a neural network-based classifier to identify the unknown grain types. In the test dataset, the classification accuracies were 90.00%, 88.00%, 95.00%, 82.00%, 74.00%, 80.00% respectively. Seeing to the above studies on rice the conclusion can be drawn that there are few quality parameters affecting the production of rice. For instance area (x1), perimeter (x2), circularity (x3), sphericity (x4), shape compactness (x5) are directly affecting the quality & non uniform broken rice (y1) are inversely proportional to quality parameter of rice. (3) Bakery Products Appearance of baked products is an important quality attribute which influences the visual perceptions of customers and hence potential demands of the products. The appearance of the internal and external features contributes to the overall impression of the products quality. Consequently, the inspection of the bakery products occupies a major
role in the manufacture process as it affects the taste, texture and appearance of the products. Computer vision system has been used to measure characteristics before packing. Variety of biscuits on which quality evaluation is done is show in figure 3.3
Figure 3.3 Variety of biscuit [17]
L. Hamey et al (1993) developed a prototype computer system for inspection of biscuit [17]. A system employs a monochromatic image and histogram of image to identify defect and classify a product as under backed, correctly backed and over backed [17]. J. C. H. Yeh et al (1995) [12] present the implementation of the inspection system with a hybrid neural network of self-organizing maps and feed forward neural networks. They have tested system and its grading performance on biscuit. The bake levels were evaluated and compared with trained human inspector. They found that the proposed color system with a hybrid neural network performed significantly better than the human inspector. They also found that the cross-validation technique can prevent over-training and preserve generalization of an ANN. Raji et al (2000) developed a program in FORTRAN using the principle of edge detection in image analysis to determine the edge of sliced breads and biscuits (round and rectangular) with a view to detecting defects (breakage) [27]. The crispness of biscuit measured in acoustic sound and not in texture of biscuit. In crispness of biscuit or snacks the distraction could happen due to environment. So, this new initial method is purposed to analyze the texture to determine the crispness of biscuit. Here they have take care of all samples of the same size and thickness because the humidity can influence if these two criteria are different. The normal texture stored as a reference to the irregular texture image with lightning condition. They applied wavelet transforms in this analyzing together with image processing from machine vision. All the texture stored in a database as a comparison to determine the good quality of biscuit [22]. A prototype-automated system for visual inspection of muffins was developed by Zaid Abdullah et al. [36]. The color of 100 light brown and 100 dark brown muffins was evaluated using the vision system and discriminants analysis compared with visual examination. The automated system was able to correctly classify 96% of pregraded and 79% of ungraded muffins. [1] For bulk sugary products classification and identification developed by B. S. Anami et al (2009). [2]. They used gray level co-occurrence method for texture analysis and feature extraction and developed a neural network model to classify 10 different varieties of bulk sugary food products. In the classification of Indian food products like Apple cake, Bundeladu, Burfi, Doodhpeda, Jamun, Jilebi, Kalakand, Ladakiladu, Mysorepak, Suraliholige, they got accuracy of 90%. Seeing to the above studies on bakery biscuits the conclusion can be drawn that there are few quality parameters affecting the baked biscuit. For instance area (x1), crispness (x2), thickness (x3) directly affect the quality & color variation (y1), less amount of ingredients (like amount of chocolate chips in chocolate cookies) (y2) are inversely proportional to quality of baked product. (4) Fruits and Vegetables Computer vision has been widely used for the quality inspection and grading of fruits and vegetables. The fresh fruit and vegetable postharvest sector is dynamic and largely to increasing consumer demand for quality product. The condition of fruit and vegetable at the time of harvest has an important effect on the consumer’s level of satisfaction
at consumption. Computer vision offers the potential to automate manual grading practices and thus to Standardize techniques and eliminate tedious inspection tasks. Color provides information about estimating the maturity and also in examining the freshness of fruits & vegetables [18]. P. Sudhakara Rao et al, uses HSI for color representation, which provide an efficient method for color discrimination. With the help of median density of Hue as a grading criterion, the image processing system achieved around 98 % accuracy in color inspection of Apples. They proposed new method namely Improved Radius signature [31]. This method can be effectively used for comparing many samples against a reference shape for the purpose of sorting and grading using correlation coefficient technique, Graphical analysis or Fourier transformation technique. Du and Sun (2004), shape features can be measured independently (for example, by Fourier descriptors of the planar image boundary, invariant moments) or by combining size measurements (for example, circularity, aspect ratio, compactness, eccentricity, roundness). So with, the determination of fruit size parameters allows simple shape sorting [6].
Figure 3.4 Pictures of Apples
Color is the important parameter regarding freshness of fruits and vegetables. H. M. W. Bunnik et al developed a system which quantifies the color of entire object using mean, standard deviation, skew and kurtosis [19]. Seeing to the above studies on fruits & vegetables the conclusion can be drawn that there are few quality parameters affecting the quality. For instance area (x1), shape (x2), size (x3), color (x4), shape compactness (x5) are directly affecting the quality & black moles (y1), brown undubbed shapes (y2) are inversely proportional to quality parameter of fruits & vegetables. (5) Fennel & Cummin seeds The booming global spice market also possesses good opportunities for the Indian spice industry especially from Gujarat (Unjha) and Tamil Nadu to provide quality spices at competitive prices. Unjha faces stiff competition from China, Malaysia and Pakistan in terms of pricing of the products. Therefore the need of the manufacturers there is to ensure consistency in supply, product quality, pricing and marketing strategy to increase the share in exports. The major one out of all is of quality [14][15].
Figure 3.5 figures of cumin & fennel seeds [14][15]
K. Jain et al [14] [15] proposed to use machine vision techniques for quality evaluation of cummin and fennel seeds. The approach was based on finding minor axis length, major axis length and area of the seeds. The classification of good and bad quality was done on finding the number of seeds with pedestals and foreign elements available in bulk of samples as shown in figure 3.5. Here, the quality was thought to be inversely proportional to number of seeds with pedestals (x1) and number of foreign elements (x2) present in the samples. 4
PROPOSED UNIFIED APPROACH FOR QUALITY EVALUATION
General review in the field of quality depicts that there is a requirement of a unified approach for the quality evaluation of food products which widely range from raw to processed product. The old rule of thumb says that acquire an image, pre process it, segment it, then extract the features and finally find its quality based on those
features by comparing it with the standard one. The features to be extracted would vary for each and every food product as discussed above. So it can be concluded in general that those features would affect the quality of specific food product. In general, some features (xi) will be directly proportional to the quality of the food product while other features (yi) will be indirectly proportional to the quality. Hence a general formula of quality of any food product based on machine vision extracted features can be given by n
c1 ∑ wi xi Q=
i =1 m
…………………………………..(1)
c2 ∑ w j y j j =1
Where, wi and wj are weights for features xi and yj respectively. The weights can be set as per the food quality requirement. The constants c1 and c2 plays a major role in defining final grades from the quality. Example of grading can be found in [14], [15]. 5
CONCLUSION
In this paper various food products quality evaluation is discussed in detail. For every product features are figured out for which the quality is either directly or inversely varied. A formula of quality is devised out which try to generate a unification approach for quality evaluation of all types of food products. The weighted sum of features based formula suggests to use proper weights for each and every features on which the quality is dependent on. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]
Abdullah, Z. M., Aziz, A. S., & Dos-Mohamed, A. M. Quality inspection of bakery products using a color-based machine vision system. Journal of Food Quality, 23(1), 39–50. (2000). Basavaraj .S. Anami, Vishwanath. C.Burkpalli. Texture based Identification and Classification of Bulk Sugary Food Objects. ICGSTGVIP Journal, ISSN: 1687-398X, Volume 9, Issue 4: 9-14. (2009). Chen Hong, Zeng Chuanhua, Ding Youchun.Peanut Shape Recognition Based on Fourier Descriptor. 978-1-4244-3894-5/09/ IEEE(2009). CHENG Fang, LIU Zhao-yan, YING Yi-bin. Machine Vision Analysis of Characteristics and Image Information Base Construction for Hybrid Rice Seed. Rice Science, Vol. 12, No. 1: 13–18.( 2005) D. M. Hobson, R. M. Carter, Y. Yan. Characterization and Identification of Rice Grains through Digital Image Analysis. Instrumentation and Measurement Technology Conference - IMTC Warsaw, Poland, May 1-3. (IEEE) (2007) Du, C.-J. and D.-W. Sun. "Recent developments in the applications of image processing techniques for food quality evaluation." Trends in Food Science & Technology 15: 230–249. (2004). Du C-J, Sun D-W ”Learning techniques used in computer vision for food quality evaluation: a review.” Journal of Food Engineering, 72 (1), 39-55.(2006). Emadzadeh1, S.M.A. Razavi1*, and R. Farahmandfar2. Monitoring geometric characteristics of rice during processing by image analysis system and micrometer measurement. Institute of Agro physics, Polish Academy of Sciences 24: 21-27.(2010) Graves, Mark, Batchelor, Bruce G. (Bruce Godfrey), Machine Vision for the Inspection of Natural Products. 1943-670.2’8563, ISBN 1852335254. Springer-Verlag London Berlin Heidelberg. (2003). HAN Zhong-zhi, ZHAO You-gang. Quality Grade-Testing of Peanut Based on Image Processing Second International Conference on Information and Computing Science. (IEEE) pp: 21-24.(2009). James I. Davidson, Jr., Paul D. Blankenshipl, and Victor Chew. Probability Distributions of Peanut Seed Size. Peanut Science 5, 9196.(1978). Jeffrey C.H. Yeh, Leonard G.C. Hamey, Tas WestcottS and Samuel K.Y. Sung. Color Bake Inspection System Using Hybrid Artificial Neural Networks. IEEE 0-7803-2768-3/95.(1995). Journal of Scientific Research and Development”. Vol. VII. Invited Overview. February 2005. Kavindra Jain, Chintan K. Modi, Kunal Pithadiya.Non Destructive quality evaluation in spice industry with specific reference to Cuminum Cyminum L (Cumin) seeds. International Conference on Innovations & Industrial Applications, Malaysia, (IEEE) 25-26 July (2009). Kavindra Jain, Chintan K. Modi, Jalpa J Patel, 2009. Non Destructive quality evaluation in spice industry with specific reference to Cuminum Cyminum L (Cumin) seeds. International Conference on Emerging Trends in Engineering and Technology, ICETET, Nagpur (IEEE Computer Society).(2009). Kyu Hong Chai, Kang-Jin Lee, Giyoung Kim. Non-destructive Quality Evaluation Technology for Fruits and Vegetables Using NearInfrared Spectroscopy. Leognard G. C. Hamey, Annesely J. Watson, C. Tasman Westcott. Machine inspection for biscuit bake. Appears in Proc. Digital Image Computing: Techniques and Applications Sydney, 8-10 December, pages 124-129.(1993) Malamasa, E. N., et al. "A survey on industrial vision systems, applications and tools." Image and Vision Computing 21: 171–188. (2003). M. W. Bunnik, D. G. Bailey, A. J. Mawson. Objective color measurement of tomatoes and lime. M. Hobson, R. M. Carter, Y. Yan. Characterization and Identification of Rice Grains through Digital Image Analysis, Instrumentation and Measurement Technology Conference – IMTC, IEEE 1-4244-0589-0/07. (2007). Narendra V G, Hareesh K S. Prospects of Computer Vision Automated Grading and Sorting Systems in Agricultural and Food Products for Quality Evaluation. International Journal of Computer Applications (0975 – 8887) Volume 1 – No. 4.(2010)
[22] Nur Shuhada Eilani bt. Ahmad, Riza Sulaiman. Texture Analysis for Biscuit Using Wavelet. International Conference on Electrical Engineering and Informatics. 5-7 August, Selangor, Malaysia. 978-1-4244-4913-2/2009 IEEE(2009)
[23] P. Sudhakara Rao, A. Gopal, R. Revathy and K. Meenakshi. Color Analysis of Fruits Using Machine Vision System For Automatic Sorting And Grading. J. Instrum. Soc. India 34 (4) 284-291.
[24] Process Operations in Nigeria. Agricultural Engineering International: the CIGR Journal of Scientific Research and Development. Vol. VII. Invited Overview. Pp. 1-12.
[25] P. Mereda, J. Ortiz-canavate, F. J. Garcia-Ramos, M. Ruiz-Altisent. Nondestructive technologies for fruits and vegetable size determination- a review.
[26] Raji and A. Alamutu. Prospects of Computer Vision Automated Sorting Systems in Agricultural Process Operations in Nigeria. Agricultural Engineering International: the CIGR. Journal of Scientific Research and Development”. Vol. VII. (2005).
[27] Raji, A.O., Fagboun, A.A., and M.K. Dania. An Approach to Detecting Defects in Food Products. Proceedings of the First International Conference of the Nigerian Institution of Agricultural Engineers, pp.36-39. (2000).
[28] S.Arivazhagan, R.Newlin Shebiah, S.Selva Nidhyanandhan, L.Ganesa.. Fruit Recognition using Color and Texture Features Journal of Emerging Trends in Computing and Information Sciences, E‐ISSN 2218‐6301. VOL. 1, NO. 2: 90-94. (2010).
[29] Sanjivani Shantaiya, Mrs.Uzma Ansari. Identification Of Food Grains And Its Quality Using Pattern Classification. International Conference [ICCT], 3rd-5th December. Special Issue of IJCCT Vol. 2 Issue 2, 3, 4pp : 70-74. (2010).
[30] Siriluk Sansomboonsuk Nitin Afzulpurkar. Machine vision for Rice Quality Evaluation. Technology and Innovation for Sustainable Development Conference (TISD) pp: 343-346. (2008)
[31] Sudhakara Rao, A. Gopal, S. Md. Iqbal, R. Revathy and K. Meenakshi Classification of Fruits Based on Shape Using Image-Processing Techniques. J. Instrum. Soc. India 34 (4) pp: 227-239.
[32] Tadhg Brosnan, Da-Wen Sun. Improving quality inspection of food products by computer vision-a review, Journal of Food Engineering 61,pp. 3–16.(2004)
[33] Wan, Y.-N. Kernel handling performance of an automatic grain quality inspection system. Trans. ASAE 45(2) pp:369-378. (2002). [34] Wan, Y.-N., C.-M. Lin, and J.-F. Chiou. Rice quality classification using an automatic grain quality inspection system. Trans. ASAE 45(2) pp: 379-388. (2002)
[35] Zheng C, Sun D-W, Zheng L. “Recent applications of image texture for evaluation of food qualities - a review.” Trends in Food Science and Technology, 17(3), pp: 113-128. (2006).
[36] Zaid Abdullah, M., Abdul Aziz, S., Dos- Mohamed, AM. Quality inspection of bakery products using a color-based machine vision system. Journal of Food Quality, 23(1), pp.39-50. (2000).
Abstract— The paper presents a unified approach for quality evaluation of food using image processing and machine vision. In this paper basic tool is combination of computer and machine vision for image analysis and processing through which fast and accurate quality is achieved that too with the help of non-destructive method. Machine vision in food has broadened its range of applications from grains, cereals, fruits to vegetables including processed products as well as spices in which there is a high degree of quality achieved as compared to human vision inspection. In this paper we quantify the qualities of various food products and figure out features which are directly or inversely affect the quality of the food product. Based on these features a generalized formula of quality is proposed to be used for quality evaluation of any type of food product. Keywords: Quality, Machine vision, Image processing, Fruit, Vegetables, Grains, Spices, Unification approach.
1.
INTRODUCTION
The agricultural industry is probably too oldest and most widespread industry in the world. Quality control is of major importance in the food industry. After harvesting a food product based on quality parameter a food product has been sorted and graded in different grades. Also a continuous development in mechanical harvesting system and need for automatic grading system has arisen. In recent years due to rapid industrialization and massive rural-urban migration the necessity of agricultural mechanization needs to be grown faster. Non-destructive quality evaluation of food products is an important and very vital factor in food/agricultural industry. Various parameters which define quality of these products (e.g. size, shape, color, texture, external defects, smell etc.) are evaluated by human inspectors [35]. Together with the high labor costs, inconsistency and variability associated with human inspection accentuates the need for objective measurements systems. Efforts are being geared towards the replacement of human operator with automated systems, as human operations are inconsistent and less efficient [14]. Rapid advances in hardware and software for digital image processing have motivated several studies on the development of computer vision systems to evaluate quality of diverse raw and processed foods. Advancement in computer technology leads to use these in the domain of food processing like grading, sorting and quality inspection [32]. Computer vision systems are recognized as the integration of devices for non-contact optical sensing, and computing and decision processes which receive and interpret automatically an image of a real scene. Recently many different features like size, shape, color and texture of food products are combined together for their applications in the food industry. Normally, by increasing the features used, the performance of the methods proposed can be increased. However, in literature there is a need of a common evaluation scheme. In this paper we propose a unification approach to evaluate the quality of food products. In this paper we present a unified approach to evaluate the quality of food products using machine vision. Section II discusses basic machine vision technique for quality evaluation of food. Section III discusses background of various food products in which machine vision is used. We propose a unified approach for evaluation of quality of food products in section IV. Section V concludes the paper. 2.
QUALITY EVALUATION USING MACHINE VISION
Computer & Machine vision system for quality evaluation of food products contains standard hardware configuration as shown in figure 2.1. It consists of : 1. Some means of presenting the object to be inspected to the camera
2. 3. 4. 5.
Light source for proper illumination CCD Camera to acquire image A frame-grabber, to perform analog-to-digital conversion. Software, if a computer is used for image processing
Figure 2.1 Imaging setup of computer vision system [5]
Illumination plays a vital role in quality of image. So here proper illumination is adjusted with the help of available light sources. Image acquisition is done (or image formation) to produce digital images of the food products. Image is captured by high quality CCD camera. Pre-processing is used to enhance, clean, and improve key parts of the input images [15], [16]. During the process of image acquisition, captured images may suffer from severe noise and external disturbances this can be removed with the help of filtering process. Image can be converted from color to gray scale depending upon whether color parameter is of importance or not. Segmentation is done to identify relevant parts of a food product, as disjoint and non-overlapping regions separated from the background. To partition an image, thresholding is applied on image. Object recognition and measurement are done to quantify key features of the selected objects. Here different parameters based on size, shape etc. is calculated. Edge operators detect discontinuities in grey level, color, texture, etc. Region segmentation involves the grouping together of similar pixels to form regions representing single objects within the image. In interpretation, where the features extracted from the products are interpreted using application domain knowledge. 3
ASSESMENT OF FOOD PRODUCTS
Food product images captured by vision system can be used to identify, analyze and quality assessment of food products. (1) Peanuts Peanut is a mass consumption item and is used for extraction of oil, for making butter, chikkies and chocolates, as an ingredient in making several food and various snacks. India is one of the largest producers along with the USA; China and Argentina. Gujarat, Andhra Pradesh, Tamilnadu and Maharashtra are the main cultivating states. Peanuts from the Saurashtra region of Gujarat are famous all over the world on account of their big size, nutty flavor and crunchy taste. Peanuts having a wide range of seed size and maturity are often obtained at harvest. Due to nonuniform maturity levels size variation has been seen in peanuts like a smaller sized seed would be immature seed and a big seed would be mature as shown in figure 3.1 [3] [11].
Figure 3.1 Partial pictures of peanut in different varieties [11]
In 1978, J. I. Davidson derived a mathematical relationship for seed size distribution to describe relation between different seed (arachies hypogaea L.) varieties. A step further in 1989, F. E. Dowell, who developed an automatic peanut grading system based on machine vision system with the integration of mechanical component to grade peanuts based on size or defect [11]. In 2009, H. Zhong-zhi and Z. You-gang compared 48 different peanut varieties. Using principal component analysis they developed an artificial neural network to establish a seed recognition model which was made up of 49 distinct appearance characteristics referred as shape, texture, color etc.. Here they got variety recognition rate and quality recognition rate up to 91.2% and 93.0% respectively [10]. In 2010, H. Zhong-zhi et al developed a neural network, which was based on 52 appearance features. They got success up to 95.6% in their results. This method has been used for appraisal of peanut quality in China. Seeing to the above studies on peanut the conclusion can be drawn that there are few quality parameters affecting the production of peanuts. For instance color (x1), texture (x2) are directly affecting the quality & non uniform maturity level i.e. size (major (y1) and minor (y2) axis) are inversely proportional to quality of peanuts. (2) Rice Rice is one of the leading food crops of the world and more than half of the world’s population relies on rice as the major daily source of calories and protein [4]. The quality of rice has distinct effect, so the proper inspection of rice quality is very important. There are two main factors for checking the quality of rice kernel: percentage of broken rice and percentage of the purity of rice as shown in figure 3.2.
Figure 3.2 Picture of rice characteristics [5]
In 2002, Y. N. Wan developed an automatic grain kernel handling system in which 1296 singularized kernel images per minute were taken for machine vision inspection. He developed a Windows-based software program for rice quality inspection. [33]. Later on S, Sansomboonsuk and N. Afzulpurkar (2008) developed image processing algorithms and used to extract features for kernels [30]. They measured and calculated area, perimeter, circularity and shape compactness as criteria in Fuzzy logic for classifying each kernel. From testing the image analysis algorithms, the results are obtained, they found accuracy averaging 92% for both of the broken rice and the purity of rice compared with human inspection. In 2009, B. Emadzadeh, S.M.A. Razavi, and R. Farahmandfar [9] compared three Iranian rice varieties, namely Tarom Mahalli, Fajr and Neda, determined parameters using micrometer and image processing methods. Comparison of the results obtained by both procedures showed that the geometric characteristics (length, width, height and projected area) of all three varieties decrease and sphericity increases after removing the outer and the brownish layers. It was found that the values of micrometer data are having lower for all the geometric factors and that the true size and sphericity. In 2010, S. Shantaiya, U. Ansari developed digital image analysis algorithm based on color, morphological and textural features to identify the six varieties rice seeds in Chhattisgarh region [29]. Nine color and nine morphological and textural features were used for analysis. They developed a neural network-based classifier to identify the unknown grain types. In the test dataset, the classification accuracies were 90.00%, 88.00%, 95.00%, 82.00%, 74.00%, 80.00% respectively. Seeing to the above studies on rice the conclusion can be drawn that there are few quality parameters affecting the production of rice. For instance area (x1), perimeter (x2), circularity (x3), sphericity (x4), shape compactness (x5) are directly affecting the quality & non uniform broken rice (y1) are inversely proportional to quality parameter of rice. (3) Bakery Products Appearance of baked products is an important quality attribute which influences the visual perceptions of customers and hence potential demands of the products. The appearance of the internal and external features contributes to the overall impression of the products quality. Consequently, the inspection of the bakery products occupies a major
role in the manufacture process as it affects the taste, texture and appearance of the products. Computer vision system has been used to measure characteristics before packing. Variety of biscuits on which quality evaluation is done is show in figure 3.3
Figure 3.3 Variety of biscuit [17]
L. Hamey et al (1993) developed a prototype computer system for inspection of biscuit [17]. A system employs a monochromatic image and histogram of image to identify defect and classify a product as under backed, correctly backed and over backed [17]. J. C. H. Yeh et al (1995) [12] present the implementation of the inspection system with a hybrid neural network of self-organizing maps and feed forward neural networks. They have tested system and its grading performance on biscuit. The bake levels were evaluated and compared with trained human inspector. They found that the proposed color system with a hybrid neural network performed significantly better than the human inspector. They also found that the cross-validation technique can prevent over-training and preserve generalization of an ANN. Raji et al (2000) developed a program in FORTRAN using the principle of edge detection in image analysis to determine the edge of sliced breads and biscuits (round and rectangular) with a view to detecting defects (breakage) [27]. The crispness of biscuit measured in acoustic sound and not in texture of biscuit. In crispness of biscuit or snacks the distraction could happen due to environment. So, this new initial method is purposed to analyze the texture to determine the crispness of biscuit. Here they have take care of all samples of the same size and thickness because the humidity can influence if these two criteria are different. The normal texture stored as a reference to the irregular texture image with lightning condition. They applied wavelet transforms in this analyzing together with image processing from machine vision. All the texture stored in a database as a comparison to determine the good quality of biscuit [22]. A prototype-automated system for visual inspection of muffins was developed by Zaid Abdullah et al. [36]. The color of 100 light brown and 100 dark brown muffins was evaluated using the vision system and discriminants analysis compared with visual examination. The automated system was able to correctly classify 96% of pregraded and 79% of ungraded muffins. [1] For bulk sugary products classification and identification developed by B. S. Anami et al (2009). [2]. They used gray level co-occurrence method for texture analysis and feature extraction and developed a neural network model to classify 10 different varieties of bulk sugary food products. In the classification of Indian food products like Apple cake, Bundeladu, Burfi, Doodhpeda, Jamun, Jilebi, Kalakand, Ladakiladu, Mysorepak, Suraliholige, they got accuracy of 90%. Seeing to the above studies on bakery biscuits the conclusion can be drawn that there are few quality parameters affecting the baked biscuit. For instance area (x1), crispness (x2), thickness (x3) directly affect the quality & color variation (y1), less amount of ingredients (like amount of chocolate chips in chocolate cookies) (y2) are inversely proportional to quality of baked product. (4) Fruits and Vegetables Computer vision has been widely used for the quality inspection and grading of fruits and vegetables. The fresh fruit and vegetable postharvest sector is dynamic and largely to increasing consumer demand for quality product. The condition of fruit and vegetable at the time of harvest has an important effect on the consumer’s level of satisfaction
at consumption. Computer vision offers the potential to automate manual grading practices and thus to Standardize techniques and eliminate tedious inspection tasks. Color provides information about estimating the maturity and also in examining the freshness of fruits & vegetables [18]. P. Sudhakara Rao et al, uses HSI for color representation, which provide an efficient method for color discrimination. With the help of median density of Hue as a grading criterion, the image processing system achieved around 98 % accuracy in color inspection of Apples. They proposed new method namely Improved Radius signature [31]. This method can be effectively used for comparing many samples against a reference shape for the purpose of sorting and grading using correlation coefficient technique, Graphical analysis or Fourier transformation technique. Du and Sun (2004), shape features can be measured independently (for example, by Fourier descriptors of the planar image boundary, invariant moments) or by combining size measurements (for example, circularity, aspect ratio, compactness, eccentricity, roundness). So with, the determination of fruit size parameters allows simple shape sorting [6].
Figure 3.4 Pictures of Apples
Color is the important parameter regarding freshness of fruits and vegetables. H. M. W. Bunnik et al developed a system which quantifies the color of entire object using mean, standard deviation, skew and kurtosis [19]. Seeing to the above studies on fruits & vegetables the conclusion can be drawn that there are few quality parameters affecting the quality. For instance area (x1), shape (x2), size (x3), color (x4), shape compactness (x5) are directly affecting the quality & black moles (y1), brown undubbed shapes (y2) are inversely proportional to quality parameter of fruits & vegetables. (5) Fennel & Cummin seeds The booming global spice market also possesses good opportunities for the Indian spice industry especially from Gujarat (Unjha) and Tamil Nadu to provide quality spices at competitive prices. Unjha faces stiff competition from China, Malaysia and Pakistan in terms of pricing of the products. Therefore the need of the manufacturers there is to ensure consistency in supply, product quality, pricing and marketing strategy to increase the share in exports. The major one out of all is of quality [14][15].
Figure 3.5 figures of cumin & fennel seeds [14][15]
K. Jain et al [14] [15] proposed to use machine vision techniques for quality evaluation of cummin and fennel seeds. The approach was based on finding minor axis length, major axis length and area of the seeds. The classification of good and bad quality was done on finding the number of seeds with pedestals and foreign elements available in bulk of samples as shown in figure 3.5. Here, the quality was thought to be inversely proportional to number of seeds with pedestals (x1) and number of foreign elements (x2) present in the samples. 4
PROPOSED UNIFIED APPROACH FOR QUALITY EVALUATION
General review in the field of quality depicts that there is a requirement of a unified approach for the quality evaluation of food products which widely range from raw to processed product. The old rule of thumb says that acquire an image, pre process it, segment it, then extract the features and finally find its quality based on those
features by comparing it with the standard one. The features to be extracted would vary for each and every food product as discussed above. So it can be concluded in general that those features would affect the quality of specific food product. In general, some features (xi) will be directly proportional to the quality of the food product while other features (yi) will be indirectly proportional to the quality. Hence a general formula of quality of any food product based on machine vision extracted features can be given by n
c1 ∑ wi xi Q=
i =1 m
…………………………………..(1)
c2 ∑ w j y j j =1
Where, wi and wj are weights for features xi and yj respectively. The weights can be set as per the food quality requirement. The constants c1 and c2 plays a major role in defining final grades from the quality. Example of grading can be found in [14], [15]. 5
CONCLUSION
In this paper various food products quality evaluation is discussed in detail. For every product features are figured out for which the quality is either directly or inversely varied. A formula of quality is devised out which try to generate a unification approach for quality evaluation of all types of food products. The weighted sum of features based formula suggests to use proper weights for each and every features on which the quality is dependent on. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]
Abdullah, Z. M., Aziz, A. S., & Dos-Mohamed, A. M. Quality inspection of bakery products using a color-based machine vision system. Journal of Food Quality, 23(1), 39–50. (2000). Basavaraj .S. Anami, Vishwanath. C.Burkpalli. Texture based Identification and Classification of Bulk Sugary Food Objects. ICGSTGVIP Journal, ISSN: 1687-398X, Volume 9, Issue 4: 9-14. (2009). Chen Hong, Zeng Chuanhua, Ding Youchun.Peanut Shape Recognition Based on Fourier Descriptor. 978-1-4244-3894-5/09/ IEEE(2009). CHENG Fang, LIU Zhao-yan, YING Yi-bin. Machine Vision Analysis of Characteristics and Image Information Base Construction for Hybrid Rice Seed. Rice Science, Vol. 12, No. 1: 13–18.( 2005) D. M. Hobson, R. M. Carter, Y. Yan. Characterization and Identification of Rice Grains through Digital Image Analysis. Instrumentation and Measurement Technology Conference - IMTC Warsaw, Poland, May 1-3. (IEEE) (2007) Du, C.-J. and D.-W. Sun. "Recent developments in the applications of image processing techniques for food quality evaluation." Trends in Food Science & Technology 15: 230–249. (2004). Du C-J, Sun D-W ”Learning techniques used in computer vision for food quality evaluation: a review.” Journal of Food Engineering, 72 (1), 39-55.(2006). Emadzadeh1, S.M.A. Razavi1*, and R. Farahmandfar2. Monitoring geometric characteristics of rice during processing by image analysis system and micrometer measurement. Institute of Agro physics, Polish Academy of Sciences 24: 21-27.(2010) Graves, Mark, Batchelor, Bruce G. (Bruce Godfrey), Machine Vision for the Inspection of Natural Products. 1943-670.2’8563, ISBN 1852335254. Springer-Verlag London Berlin Heidelberg. (2003). HAN Zhong-zhi, ZHAO You-gang. Quality Grade-Testing of Peanut Based on Image Processing Second International Conference on Information and Computing Science. (IEEE) pp: 21-24.(2009). James I. Davidson, Jr., Paul D. Blankenshipl, and Victor Chew. Probability Distributions of Peanut Seed Size. Peanut Science 5, 9196.(1978). Jeffrey C.H. Yeh, Leonard G.C. Hamey, Tas WestcottS and Samuel K.Y. Sung. Color Bake Inspection System Using Hybrid Artificial Neural Networks. IEEE 0-7803-2768-3/95.(1995). Journal of Scientific Research and Development”. Vol. VII. Invited Overview. February 2005. Kavindra Jain, Chintan K. Modi, Kunal Pithadiya.Non Destructive quality evaluation in spice industry with specific reference to Cuminum Cyminum L (Cumin) seeds. International Conference on Innovations & Industrial Applications, Malaysia, (IEEE) 25-26 July (2009). Kavindra Jain, Chintan K. Modi, Jalpa J Patel, 2009. Non Destructive quality evaluation in spice industry with specific reference to Cuminum Cyminum L (Cumin) seeds. International Conference on Emerging Trends in Engineering and Technology, ICETET, Nagpur (IEEE Computer Society).(2009). Kyu Hong Chai, Kang-Jin Lee, Giyoung Kim. Non-destructive Quality Evaluation Technology for Fruits and Vegetables Using NearInfrared Spectroscopy. Leognard G. C. Hamey, Annesely J. Watson, C. Tasman Westcott. Machine inspection for biscuit bake. Appears in Proc. Digital Image Computing: Techniques and Applications Sydney, 8-10 December, pages 124-129.(1993) Malamasa, E. N., et al. "A survey on industrial vision systems, applications and tools." Image and Vision Computing 21: 171–188. (2003). M. W. Bunnik, D. G. Bailey, A. J. Mawson. Objective color measurement of tomatoes and lime. M. Hobson, R. M. Carter, Y. Yan. Characterization and Identification of Rice Grains through Digital Image Analysis, Instrumentation and Measurement Technology Conference – IMTC, IEEE 1-4244-0589-0/07. (2007). Narendra V G, Hareesh K S. Prospects of Computer Vision Automated Grading and Sorting Systems in Agricultural and Food Products for Quality Evaluation. International Journal of Computer Applications (0975 – 8887) Volume 1 – No. 4.(2010)
[22] Nur Shuhada Eilani bt. Ahmad, Riza Sulaiman. Texture Analysis for Biscuit Using Wavelet. International Conference on Electrical Engineering and Informatics. 5-7 August, Selangor, Malaysia. 978-1-4244-4913-2/2009 IEEE(2009)
[23] P. Sudhakara Rao, A. Gopal, R. Revathy and K. Meenakshi. Color Analysis of Fruits Using Machine Vision System For Automatic Sorting And Grading. J. Instrum. Soc. India 34 (4) 284-291.
[24] Process Operations in Nigeria. Agricultural Engineering International: the CIGR Journal of Scientific Research and Development. Vol. VII. Invited Overview. Pp. 1-12.
[25] P. Mereda, J. Ortiz-canavate, F. J. Garcia-Ramos, M. Ruiz-Altisent. Nondestructive technologies for fruits and vegetable size determination- a review.
[26] Raji and A. Alamutu. Prospects of Computer Vision Automated Sorting Systems in Agricultural Process Operations in Nigeria. Agricultural Engineering International: the CIGR. Journal of Scientific Research and Development”. Vol. VII. (2005).
[27] Raji, A.O., Fagboun, A.A., and M.K. Dania. An Approach to Detecting Defects in Food Products. Proceedings of the First International Conference of the Nigerian Institution of Agricultural Engineers, pp.36-39. (2000).
[28] S.Arivazhagan, R.Newlin Shebiah, S.Selva Nidhyanandhan, L.Ganesa.. Fruit Recognition using Color and Texture Features Journal of Emerging Trends in Computing and Information Sciences, E‐ISSN 2218‐6301. VOL. 1, NO. 2: 90-94. (2010).
[29] Sanjivani Shantaiya, Mrs.Uzma Ansari. Identification Of Food Grains And Its Quality Using Pattern Classification. International Conference [ICCT], 3rd-5th December. Special Issue of IJCCT Vol. 2 Issue 2, 3, 4pp : 70-74. (2010).
[30] Siriluk Sansomboonsuk Nitin Afzulpurkar. Machine vision for Rice Quality Evaluation. Technology and Innovation for Sustainable Development Conference (TISD) pp: 343-346. (2008)
[31] Sudhakara Rao, A. Gopal, S. Md. Iqbal, R. Revathy and K. Meenakshi Classification of Fruits Based on Shape Using Image-Processing Techniques. J. Instrum. Soc. India 34 (4) pp: 227-239.
[32] Tadhg Brosnan, Da-Wen Sun. Improving quality inspection of food products by computer vision-a review, Journal of Food Engineering 61,pp. 3–16.(2004)
[33] Wan, Y.-N. Kernel handling performance of an automatic grain quality inspection system. Trans. ASAE 45(2) pp:369-378. (2002). [34] Wan, Y.-N., C.-M. Lin, and J.-F. Chiou. Rice quality classification using an automatic grain quality inspection system. Trans. ASAE 45(2) pp: 379-388. (2002)
[35] Zheng C, Sun D-W, Zheng L. “Recent applications of image texture for evaluation of food qualities - a review.” Trends in Food Science and Technology, 17(3), pp: 113-128. (2006).
[36] Zaid Abdullah, M., Abdul Aziz, S., Dos- Mohamed, AM. Quality inspection of bakery products using a color-based machine vision system. Journal of Food Quality, 23(1), pp.39-50. (2000).
