Digital Inspection in Food Industry - Essay Example

? Digital inspection in food industry Table of contents 0 Introduction 3 2.0 Research Questions 3 3.0 Research Methodology 4 4.0 Limitations of research 4 5.0 Literature Review 4 5.1 Digital inspection in food industry – overview 4 5.2 Components of the system involved in computer vision 7 5.3 Computer vision – phases 8 5.3.1 Image processing 8 5.3.2 Image analysis 9 5.4 Evaluation and risks 10 6.0 Conclusion 12 7.0 Issues for further research 13 8.0 References 13 1.0 Introduction The use of digital devices for checking the quality of food is a popular practice in countries worldwide. Despite its value, as verified by scientists and researchers, the application of this practice has been related to a series of problems. More specifically, it seems that not all organizations are able to support the development of the relevant techniques either because the lack of funds or because of the lack of appropriately skilled staff. Current paper explores the various aspects of foods’ digital inspection, which is commonly developed through computer vision. The requirements of this process are analyzed and its value is explained. Examples, as available in the literature, are presented so that the benefits and the risks of the specific technique are clear. It is proved that the digital inspection of foods through computer vision can be quite helpful in identifying defects in foods but also in checking the actual ingredients of foods provided to the public. This means that digital inspection, as applied using computer vision, can serve the need for testing the appropriateness of foods for consumption but also for identifying their ingredients, if they are aligned with the standards related to the particular type of food or not. 2.0 Research Questions The research questions on which this study is based can be described as follows: A) which are the key characteristics and role of digital inspection in regard to the food industry? B) Which are the components of digital inspection with computer vision? C) Which are the phases of computer vision in digital inspection of foods? D) Which are the risks related with the use of computer vision in digital inspection of foods? 3.0 Research Methodology Literature Review has been chosen as the research methodology for exploring the various aspects of this study’s subject. More specifically, a series of articles published in academic journals has been employed for answering to the study’s research questions, as described in the previous section. Effort has been made to retrieve recent studies. In fact, the studies used have been published from 2011 up today, with an exception of 3 studies, 2 of which were published in 2004 and one was published in 2006. These studies were used, even if not quite recent, at the level they analyzed issues that need to be incorporated in this paper. 4.0 Limitations of research During the research developed for this study the following issue appeared: not all issues addressed in this study are sufficiently covered in the literature. Effort has been made to refer to as many aspects of the study’s subject as possible; however, it is possible that gaps can be identified in regard to the answers given to the study’s research questions. These gaps are identified and highlighted in the end of the study making suggestions for future research. The fact that the recent studies available for the particular subject are limited should be also highlighted; older studies have been used instead for covering this gap so that the credibility of the study is not affected. 5.0 Literature Review 5.1 Digital inspection in food industry – overview and role Digital inspection of foods, as a process, is incorporated in food engineering sector (Yam and Papadakis 2004). The specific process focuses on the analysis of specific characteristics of food so that its condition and its synthesis are checked. This type of analysis is characterized as qualitative analysis and ‘involves visual inspection and comparison of the food samples’ (Yam and Papadakis 2004: 137). In its most common form, digital inspection is developed through computer vision. Computer vision is a process through which ‘the images of physical objects are used for describing the objects’ (Brosnan and Sun 2004:4). Computer vision has been proved to be particularly effective in analyzing a high range of foods, both fresh products, such as fish and vegetables and processed food products, such as bread and ready meals (Brosnan and Sun 2004:3). The key reason for the introduction of computer vision in the inspection of foods has been the following: traditionally, food inspection has been based on the examination by a sector’s specialist of food’s physical characteristics, such as ‘appearance, smell and flavour’ (Brosnan and Sun 2004:4) has been related to many failures, especially since the view of each specialist would be depended on personal beliefs, meaning that the assumption made would be based on subjective criteria. The need for the development of methods based on technology has been, therefore, emergent; computer vision in the inspection of foods is based on a series of standards and rules, leading to a subjective description of the food under evaluation. From a similar point of view, Patel et al. (2012) noted that the control over the quality of food has been a major challenge for food industries worldwide. Indeed, the introduction of strict rules in regard to food quality has led companies in the food industry to seek for quality inspection methods the cost of which would be limited, compared to the traditional methods of quality inspection which was characterized by the non-involvement of machines in the various phases of the process (Patel et al. 2012). Machine vision has been introduced for reducing the cost and the time requirements of the quality inspection of foods (Patel et al. 2012). In other words, the high advantages of machine/ computer vision as a process for checking the quality of foods are the following: ‘flexibility and repeatability’ (Patel et al. 2012, p.124). It should be noted that machine/ computer vision is highly popular in many sectors; in the food sector the specific process is related to the control of the ingredients and the surface of foods, as revealed through the table in Figure 1 below; in this table the role of machine vision in different fields, including the food industry, is analytically explained. Figure 1 – Use of machine vision in various sectors, including the food industry (source: Patel et al. 2012:125) Narendra and Hareesh (2010) note that machine vision is ‘a consistent, rapid, economic and objective inspection technique’ (p.43) being applied in a series of industries. The co-existence of the above characteristics has made the machine vision a popular technique for checking quality in regard to various products (Narendra and Hareesha 2010). In regard specifically to the food industry, machine vision can be used for checking not only ‘the quality but also the grading of foods’ (Narendra and Hareesh 2010: 43). 5.2 Components of the system involved in computer vision The structure of a common system involved in computer vision can be characterized as rather simple. A common computer vision system is presented in the graph in Figure 2 below. As made clear through the particular graph, the digital inspection through computer vision is a process that can be developed rather easily, even by a person who has basic knowledge of IT systems. Of course, advanced systems based on computer vision are also available for foods’ digital inspection, offering the advantage of more detailed information on the ingredients of foods under evaluation. Figure 2 – A computer vision system, components (source: Brosnan and Sun 2004:4) 5.3 Computer vision – phases Two are the key phases in which the food inspection using a computer vision is developed: a) a camera is used for taking photos of the food product which needs to be analyzed; the image of the product is stored in the camera so that they are available for processing; b) the process of the images stored is made using an IT system; specialized software is required for processing the image and for developing assumptions in regard to the food under examination (Khojel, Bodhe and Adsul, 2013). The above two phases of the process are developed in turn, i.e. first the image is obtained and then, at a second phase, the image is processed (Khojel, Bodhe and Adsul, 2013). 5.3.1 Image processing In order to understand the value and the role of Computer vision it is necessary to refer in detail to its phases. In the first phase, i.e. when taking a photo of the food product involved, there is no specific condition that needs to be met. Just to ensure that the photo taken will be saved in the computer system that will be used in the image processing phase, i.e. the second phase of the process. After being saved, in an appropriate file format, the photo/ image is processed using relevant software. During the image processing process a key task is developed: the image is processed as of its characteristics, such as colour and size so that faults are identified and fixed; reference is made to faults that are able to affect the actual view of a physical object, such as ‘geometric distortion, improper focus and camera motion’ (Sadegaonkar and Wagh 2013, p.1209). It should be noted that when taking the photo of the food product under evaluation different levels of resolution can be used, so that clearer assumptions are made in regard to the synthesis of the particular food product. The above issue could be understood by the following example: Singh and Kaur (2012) have explored the potential use of computer use in checking the quality of bakery products. A photo of a baker product has been taken using ‘CCD camera of resolution more than 5 mega pixel’ (Singh and Kaur 2012: 527). The photo taken has been processed so that it can show quite clearly the elements of the product’s surface (Singh and Kaur 2012). The photo taken, and its ‘thresholded version’ (Singh and Kaur 2012: 527) are presented in Figure 3. These photos have been further analyzed in order to check the quality of the particular bakery product. The implications of this analysis are presented in section 5.3.2. Figure 3 – Photo of bread, using a camera with resolution more than 5 pixel; in the right hand, the thresholded version of photo in the left (Singh and Kaur 2012: 527) 5.3.2 Image analysis Image processing is followed by image analysis, a process used for highlighting key features of an image, mostly related to the background of the object involved (Sadegaonkar and Wagh 2013). After the completion of the image processing process quantitative data in regard to the image are gathered. Then, the information retrieved in the image processing process is checked using appropriate software (Sadegaonkar and Wagh 2013. After the end of the above control process a decision can be taken in regard to the qualities/ characteristics of the physical object presented in the image under examination (Sadegaonkar and Wagh 2013. The image processing process can be complex, as indicated in the graph in Figure 4 below. Figure 4 – A fruit grading method using image processing (source: Khojel, Bodhe and Adsul, 2013, p.3252) In section 5.3.1 the use of machine/ computer vision for checking the quality of baker products was explained by referring to the study of Singh and Kaur (2012); the above researchers taken a photo, using camera with appropriate resolution, of a piece of bread and then tried to analyze this photo so that the quality characteristics of the bread are identified. Another photo, showing this piece of bread baked, was used for identifying the changes on the bread’s surface under the influence of the baking process (Singh and Kaur 2012). The digital inspection of bread using machine vision revealed that baking leads to a series of changes on bread’s quality; these changes are mostly related to the bread’s ‘color, shape, size and texture’ (Singh and Kaur 2012: 528). The effects of baking on bread’s quality could not be evaluated otherwise, at least not in such detail. 5.4 Evaluation and risks As made clear through the graph in Figure 1, computer vision, as a method for analyzing food products, requires specific equipment: a computer and a camera. Lights and a Basis for putting the product under analysis are also required, but not necessarily. Indeed, it is possible for the camera to incorporate a flash, so no additional lights are required. Also, any flat material that can keep the food position in horizontal position, such as a piece of thick paper, can be used instead of a basis, if the latter is not available. This means that, in terms of the equipment/ devices required the Computer vision process has limited demands, a fact that possibly explains its popularity. In any case, computer vision as part of the quality inspection process has been necessary in saving time when evaluating the quality of foods. Moreover, the food production, as a process, has been highly related to automated processes; the involvement of machine vision, of an automated technique, in the quality inspection process can be considered as part of the efforts for the establishment of automated processes in all phases of production (Pedreschi et al. 2006). The potential of machine vision to store data related to any of its phases is another important advantage of machine vision as a quality inspection system (Saldana et al. 2013). It should be noted that the data required for evaluating the quality of foods can be 2-dimensional (Narendra and Hareesh 2010). More specifically, in the context of computer vision systems the analysis of images related to foods can be only developed effectively if ‘2-dimensional data are retrieved; often such data is not adequate and 3-dimensional data has to be available’ (Narendra and Hareesh 2010: 56). In other words, the type of data necessary for developing the image analysis phase of computer vision can be differentiated across computer vision systems, a fact that denotes also differences in these systems’ potential to evaluate successfully the quality of foods. This means that if the researcher is not able to retrieve 3-D data, where necessary, when using the machine vision process then the credibility of the results can be doubted. Indeed, it has been proved that the results of a machine vision system are depended on the quality of the images of products under evaluation (Saldana et al. 2013). It is implied that if the images taken using the system’s camera are of low quality then the development of false assumptions in regard to the quality of foods under examination is quite possible (Mahendran et al. 2011). Another challenge that the researcher using the machine vision has to face is the following: if the process is developed in an environment where the natural light is not available, the availability of adequate artificial lighting has to be secure, otherwise the images taken would be of low quality (Narendra and Hareesh 2010). Saldana et al (2013) note that failures in the quality of images involved in the machine vision process are often difficult to be avoided. The use of cameras that can offer ‘a real-time view of the product’ (Saldana et al. 2013: 56) can help to minimize the risks involved in regard to image quality. Algorithms are often used for securing the quality of images used in the machine vision process (Saldana et al. 2013). 6.0 Conclusion - Recommendations According to the issues discussed above machine vision is unique in revealing the characteristics of food products in terms of quality. Indeed, the empirical research developed in this field has proved that machine vision can be effectively used for a high range of food products, including bakery products, meat, fish, agricultural products and so on. One of the key problems related to the use of the specific process seems to be the following: the evaluation of the quality of food products using computer vision is not based on rules or criteria that are standardized. More specifically, after checking a food product using machine vision the researcher involved in the particular activity should make his assumptions in regard to the product’s quality characteristics using a series of standards, such as, for example, the ISO standards. Of course, it could be supported that the creation of such standards is not an easy task especially if taking into consideration the range of food products globally. On the other hand, it has been proved that computer vision is related to certain disadvantages. In this context, the introduction of standards in the quality inspection of foods through machine vision could lead to the elimination of the relevant risks and could increase the effectiveness of the process. 7.0 Issues for further research The literature focusing on machine vision emphasizes on the advantages of the particular process as part of the quality inspection of products. In regard to the food industry also the potential benefits of machine vision cannot be doubted. Still, the literature review developed for this study has revealed that the above process is related to certain disadvantages; the high dependency of machine vision on the images involved in the process seems is the most critical drawback of the process. At the same time, in the food industry, the results of computer vision are of critical importance for evaluating the potentials of a food product to enter the market. Taking such decision by being based on false data can have severe implications for the health of the public. Therefore, the disadvantages of machine vision should be addressed with no delay; the development of research for identifying methods that can secure the credibility of machine vision would be quite necessary. In the food industry, in particular, such initiatives should be taken with no delay taking into consideration the continuous increase of food needs worldwide and the severe failures in foods quality in countries internationally. References Brosnan, T. and Sun, D., 2004. “Improving quality inspection of food products by computer vision––a review.” Journal of Food Engineering, 61: 3-16 Khojel, S., Bodhe, S. and Adsul, A., 2013. “Automated Skin Defect Identification System for Fruit Grading Based on Discrete Curvelet Transform.” International Journal of Engineering and Technology, 5(4): 3251 – 3256 Mahendran, R., Jayashree, G and Alagusundaram, K., 2011. Application of Computer Vision Technique on Sorting and Grading of Fruits and Vegetables. Journal of Food Process Technologies, 1-7 Narendra, V. and Hareesh, K., 2010. Quality Inspection and Grading of Agricultural and Food Products by Computer Vision- A Review. International Journal of Computer Applications, 2(1): 43-65 Patel, K., Kar, A., Jha, S. and Khan, M., 2012. Machine vision system: a tool for quality inspection of food and agricultural products. Journal of Food Science and Technology, 49(2): 123-141 Pedreschi, F., Leona, J., Mery, D. and Moyano, P., 2006. Development of a computer vision system to measure the color of potato chips. Food Research International 39, 1092–1098 Sadegaonkar, V. and Wagh, K., 2013. “Quality Inspection and Grading Of Mangoes by Computer Vision & Image Analysis.” International Journal of Engineering Research and Applications, 3(5): 1208-1212 Saldana, E., Siche, R., Huaman, R., Lujan, M., Castro, W. and Queved, R., 2013. Computer vision system in real-time for color determination on flat surface food. Scientia Agropecuaria 4: 55 – 63. Singh, J. and Kaur, M., 2012. Visual Inspection of Bakery Products by Texture Analysis using Image Processing Techniques. IOSR Journal of Engineering 2(4): 526-528 Yam, K. and Papadakis, S., 2004. “A simple digital imaging method for measuring and analyzing color of food surfaces.” Journal of Food Engineering, 61: 137-142 Read More