Fruit Quality Identification: A Critical Review Of Previous Literature On Image Processing.

Critically evaluate minimum of three previous literatures (attached) pertaining to their project topic (Identification of Fruits Quality Using Image Processing) that contain key finding of previous authors such as methodology adopted parameters used, results obtained and important conclusions along with shortcomings / gap areas. In text citation must be made without fail.

With the advances in preservation of fresh food products like fruits, vegetables etc. It is now essential parameter to obtain the advance visualization techniques for the safety of fruit and vegetables.

It requires the implementation of a set rule for individual product based on its various features called upon as a algorithms for safe keeping. This research paper represents the analysis of an age factor of Tomato based on one of the visual feature of Tomato that is based on color. The performance of the product is been analyzed based on the color of Tomato i.e Red, Green or Orange. The performance of an algorithm was analyzed and discussed by observing the various sample products. 

Project Objective

Plants are essential sources of energy and fundamental components in solving the global warming crisis. Also, agriculture has made it easier and efficient for farmers to get higher outputs from farm produce (Devalatkar & Koli, 2016, p.1).

Devalactar and Koli (2016) argue that good fruit quality determines a good variety of fruit products.  However, manual identification of quality is a significant obstacle in assessing fruit quality since it is time-consuming and it is costly. Therefore, it is essential to determine the variety of fruits using automated sorting machines that are fast and cost-effective.

• To conduct a research study on the best methods used to identify fruit quality using image processing.
• To provide relevant information on future studies that can be done to improve the identification of image quality.

Three journal articles will be reviewed, to provide solutions on the identification of fruit qualities using image processing articles, and they will provide the required data and information on the identification of quality in fruits.

  The qualitative detection of images in fruits is essential since many people have used to old methods which are expensive and complicated. The research study provides solutions to the image quality detection problem, and it offers new advanced techniques that can be used in the detection of image qualities. It offers cheaper, faster and more accurate methods of detecting the quality of fruits.

The background

The quality of fruits to determine the age factor of a fruiting specimen using MATLAB technique, which reduced labor costs and maximized accuracy. Furthermore, the method is fast compared to the manual system (Devalactar & Koli 2016).

Kalaivani, Muruganand & Periasamy (2013) state that an image processing method in automated sorting machines identifies the quality of fruits. The variety of fruits can be determined using various algorithms. The trained databases of many fruits are collected, specific ranges are proposed, and the arrays can be used to assess the quality of fruits as either good or bad.

  Devalatkar & Koli (2016) on their study on the determination of age in fruits using the Matlab software, developed a segregation system for analysis of tomatoes color. Furthermore, image processing provides an accurate result for the detection of durability in tomatoes. Also, they analyzed the desired age factor using an image processing tool. Matlab image processing technique determined the color of the image and reported whether the plant was ready for use or it could be used later. Images from a camera connected to a Personal Computer (PC) were assessed using an image acquisition toolbox in MATLAB.

They proposed various stages that could be used in developing the segregation system. In the first stage, the region of interest was identified. Next, edge detection was done then color detection and finally the comparison of RGB values with the maximum color comparison. Finally, the analysis was done based on the maximum color value (Devalatkar & Koli, 2016).

Edge detection detects the presence of an edge in an image. Moreover, it simplifies image data to minimize the quantity to be processed.  RGB is one of the formats used to identify the color of images. The input image is represented with three size matching matrices of the image format. Each pattern corresponds to a particular shade of green, red or blue. The RGB value of the input image can be obtained using the MATLAB software. The maximum content of color can be achieved (Devalatkar & Koli, 2016).

Project Scope

Naganur & Samaki (2012) did a study on the grading and sorting of fruits using an image processing technique. The process captures the fruit image which is then transmitted to the MATLAB software for the extraction of features, grading, and classification.

Calafell & Roca (2014) state that fruits quality depends on its blemishes, size, and skin color. Furthermore, they developed an image recognition system in their study to identify maturity levels of Jatropha curcas fruit. The fruit was classified into different categories comprising of two stages namely; a phase that extracts characteristics from the pattern. Next, patterns were recognized using attributes from the first task. The backpropagation diagnosis model (BPDM) was adapted to recognize matured fruits. Therefore, the quality of Jatropha curcas fruits was identified.

Patel, Jain & Voshi (2011) used enhanced multiple feature-based algorithms to identify fruit quality. A procedure for processing images is prepared for effective extraction of features to detect fruits. The system calculates weights of elements in images such as the edge, orientation, and color.

Gonzalez & Woods (2002) did a study on the efficient use of both color and features of fruit texture for recognition. The classifier of minimum distance recognized fruits based on co-occurrences and statistical functions from the wavelength transformed subbands.

  Tiger & Verma (2013) undertook a study on the recognition of apples based on the normal or the infected. The method recognizes and classifies images of apples based on the obtained features of values using a two-layered feedforward network. The toolbox supports radial basis networks, feedforward networks, and other network paradigms. The recognition of signals is performed by the ANN technique.    

A study conducted by Sharma, K. & Kauri (2015) presents a new technique for the qualitative detection of fruits. A camera captured the images and desired features such as the shape, color, or the size of the fruit are extracted from the sample image. The ANN technique checks for the quality of fruits using various fruit features. Furthermore, its detection is more effective.

Furthermore, the system has steps such as feature extraction, sorting, and grading. It combines three processes. The first step takes an image of the fruit by a regular camera with a white background. The image is then put in the Matlab software, fruit features are taken out, and a neural network analyzes the data. A fruit sample is chosen for testing from the database. Finally, the fruits results are obtained (Ghazanfari, Kusalik & Irudayaraj, 2008).

Sharma & Kaur (2015) proposed a method of fruit grading according to the features such as the area and the major axis. In the first step, the graphic user interface is initiated. The next step involves training the network in which features for proposal methodology are extracted; the section describes feature extraction. The final step is the testing step in which the user can select the fruit sample that they want to test and finally assess the example into various categories. The groups below show the result of the ANN technique; they also show the final results  

Significance of the research study

Parameters of the ANN results Final Results

One the fruit is the best quality

Two  the fruit is of medium quality

Three the fruit is of poor quality.

Rupali & Patil (2007) studied fruit quality using image processing. They developed detection and grading system which was cost effective. They proposed two choices for grading by color or size.  The first case involved sorting circularly shaped fruits according to color and classification according to size. The automated system combined three processes such as sorting according to color, grading according to size and feature extraction. The MATLAB software determined fruit sizes.

Anami, Pujari & Yakkundimath (2011) designed a system for identifying and classifying images. The image processing system had ARM9, as the central processor. Furthermore, the program detects sizes of fruits by image processing algorithms on the platform.

Njoroge et al. (2002) used image processing to design an automated grading system on the fruit’s defects. The system had six CCD cameras; each side on the top, right and left had two cameras. X-ray imaging inspects the biological defects and image processor analyzes the fruit features, such as the color, size, grade, and the shape. The system is designed from advanced designs, automatic mechanical control, and expert fabrications.

 The automated system in image processing has conveyor systems and a grading assembly that has three plates that can be used to connect peripheral devices. A regulatory camera captures pictures to obtain fruit feature. The system has many steps used to supply light. Two annular light supply lights to avoid a shadow. The background color of the image easily extracts fruit edge characteristics (Calafell & Roca, 2014).

Processing Flow

Color detection is done according to RGB values; fruits are sorted according to their sizes and color. The color of the fruit is determined by RGB values of an image from the camera and processed using the MATLAB software, and the intensity can be detected easily (Sunil, 2012).  

  Its circular diameter can determine the size of the fruit. The main factor of detecting sizes is by the extraction of edges. The canny method is the most robust edge detection method that finds an edge. It uses two thresholds to identify the strengths and the weak sides. It is more likely to detect weak edges (Sunil, 2012).  

The natural symmetry of the fruit determines its size.  The algorithm has two parts that defining the fruit's axis on an image and one that finds the center coordinates of the shape of the fruit. The center coordinates of the fruit can be found with formula after finding the points of the edge sequence.  The diameter sequence can be determined after identifying coordinates at the center of the image (Calafell & Roca 2014).

 The detected diameter of the apple classifies apples according to size.

Conclusion

Devalatkar & Koli (2016) concluded that processing of fruit images could determine age in a tomato based on one color. Furthermore, the proposed image processing method used in color identification of tomatoes does well in large and small areas such as agriculture and industries. It is also possible to develop an application prototype for a short time using the MATLAB software. Furthermore, image processing provides an accurate, consistent and reliable method to estimate the severity of diseases. Multiple images can identify the color very fast

Literature review

Rupali & Pati (2007) concluded their study by proposing a demo version system for assessing the quality of fruits. The system has been significant since it facilitated the large-scale production of cameras. Also, conveyor systems have been modified. The study provides new integrated techniques for grading and sorting fruits. Image capture is always a challenge because of the external lighting conditions. Therefore, it uses a grayscale image that is important in determining the size of fruit, and it is not affected by external changes in the environment.

The authors proposed a method to determine fruit quality without any damage. Furthermore, the image processing technologies and computer software have made it easy to identify the variety of fruits using vision detecting technology which is cheap and fast (Ali et al. 2012).

2.5: Future Work to be carried out

According to Devalatkar & Koli, 2016, an image processing system can be optimized with two or more colors analysis of fruits.

Sharma & Kaur (2015) propose that in future, quality detection of fruits using the ANN technique should be compared with other automated and, mechanical means.

Rupali, S. & Patil, S. (2007) on fruit quality using image processing argue that collecting fruits from conveyor systems by a central plate leads to a variation in the weight of fruit. Further designs can be modified to enable an accurate collection of weight in fruits.  

References.

Sharma, K. & Kaur, M. (2015) Quality Detection of Fruits by using ANN technique. Journal of Electronic and Communication Engineering, 10(4), 35-41.

Rupali, S. & Patil, S. (2007) A fruit Quality Management System Based on Image Processing. Journal of Electronics and Communications Engineering, 8(6), 1-5.

Devalatkar, P. & Koli, S. (2016) Identification of Age Factor of Fruit (Tomato) using Matlab-Image Processing. International Journal of Recent Friends in Engineering and Research, 2(7), 1-13.

Kalaivani, R., Muruganand, S. & Periasamy, A. (2013) Identifying the quality of tomatoes in image processing using MATLAB. ISO, 2(8).

Ali, M., Ani, A., Eamus, D. & Daniel, K. (2012) Anew image-based processing technique to determine chlorophyll in plants. American Eurasian Journal of Agriculture and the Environment, 12(10), 1323-1328.

 Anami, B. Pujari, D. & Yakkundimath, R. (2011) Identification and Classification of Normal and Affected Agriculture Produce based on combined color and texture feature extraction. International Journal of Recent Trends in Engineering and Research 1(3),

Calafell, D & Roca, J. (2014) Application of Image Processing methodologies for fruit detection and analysis. International Journal of Recent Trends in Engineering and Research 4(6), 1.16.

Naganur, J. & Samaki, S. (2012) Fruits Sorting and Grading using Fuzzy Logic. International Journal of Advanced Research in Computer Engineering and Technology, 1(6).

Sunil, N. (2012).  Fruit sorting and Grading based on color and size. International Journal of Emerging Technologies in Computational and Applied Sciences, 1(7), 12-333.

Njoroge, J., Ninomiyo, K., Kondo, N & Toita, I. (2002) Automated fruit Grading using Image processing. The Society of Instrument and Control Engineers, 3(5), 1346-1351.

Ghazanfari, A., Kusalik, A. & Irudayaraj, J. (2008) Application of Multi Structure Neural Network to Sorting Pitashio Nuts. International Journal of the Neural Systems, 8(1), 55-61.

Gonzalez, R. & Woods, E. (2002) Digital Imaging Processing. London, Pearson Education. Naoshi, K., (2003) Fruit Grading Robot. International Conference on Advanced Mechatronics, 1366-1371.

 Tiger, B. & Verma, T. (2013) Identification and Classification of Normal and Infected Apples using Neural Network. International Journal of Science and Research, 2(6), 23-40.

Patel, H., Jain, K. & Voshi, M. (2011) Fruit  Detection using improved multiple features based algorithm. International Journal of computer applications. 13(2), 12-56.