Automated Mango Fruit Grading System Using Fuzzy Logic

This paper concentrates on the size of mango fruit. Mangoes grading by humans in current agricultural industry are subjective, inconsistent and inefficient because there is an individual difference in visual inspecting which is affected by environment, physical and psychological conditions. In this paper, fuzzy logic is used to create a novel grading method. A membership function and fuzzy rules are generated from training instances based on minimum entropy formulas. Computer and Red Green and Blue (RGB) fiber optic sensor are used to examine and clarify data corresponding to human judgment and intelligence. A total of 77.78% of accuracy is achevied under the proposed method which capable of differentiating three different grades of mango. This paper offers a competent practice and capable to be applied to improve and standardize the current mango fruit grading system.


Introduction
Mango fruit (Mangifera Indica) is native to South Asia.Thailand is the main producer of the mangoes in South East Asia with production of 2.55 million tonnes in 2010-11 (Statistical Division, FAO, 2013).Other than Thailand, there are also other countries that are supplying mangoes to the world such as Indonesia, Australia and Philippines.According to Department of Statistics Malaysia, production of mango in Malaysia in 2011 is 21,021.5 tonnes with an estimation value of 111 million ringgit (Fruit Crops Statistic, Department of Agriculture Peninsular Malaysia, 2011).Mango is a tropical fruit that has high demand every year and also gets a good price in the world market.It is important to select the fruit with the right degree of maturation in order to ensure the supply of high-quality fruits.
Over the past 20 years, fruit categorizations in agriculture industry have rapidly changed from traditional human grading system to automated grading system.In order to improve agriculture sector, implementation of Information and Communication Technology (ICT) was proposed and the idea of using ICT in agriculture sector initiates the development of an automated grading system (Mustafa et al., 2009).Several features such as size, appearance and flavor are the key indicators to determine the grade or quality of the agriculture products.In agriculture industry, many companies are using automated grading technique in fruits such as peaches and oranges (Blasco et al., 2007).Other fruit such as apple also using automated grading technique (Kavdir & Guyer, 2003).Different fruits have their different attributes to be considered when they go through the grading process.Size, color, stem location and detection of external blemishes are the chosen characteristics for oranges, peaches and apples (Blasco et al., 2003).Size, color and shape have been considered in the grading process of tomato (Jahns et al., 2001) and strawberry (Liming & Yanchao, 2010).
In Malaysia, the present mango grading process is by using the human expert to grade mangoes.Human expert means that the grading process is done by eyes and hands of the experts.This method creates some disadvantages such as lack of objectivity, accuracy and efficiency.In this study, we applied fuzzy technique in order to determine the size of the mangoes based on RGB digital fiberoptic sensor and hence graded them accordingly.

Materials and Methods
This proposed grading system relies on the sizes of the mangoes.Thus, size measurement plays a vital position in creating this grading system.There are seven processing level involved in this study.The flow of each processing level is summarized in Figure 1.

Results
The develo an appropr the RGB v   It is clear that fuzzy logic grading system has overcome traditional human grading system in term of accuracy and efficiency in grading the mango because the proposed system accomplished a 77.78% of accuracy in all three categories.

Conclusion
This research has established that by implementing fuzzy logic together with several computer software and minimum entropy formulas, the accuracy of mango grading is high.This technique begins with taking the data of mango and inserting them into the program and fuzzy logic technique was implemented in grading process.The mango is categorized based on the RGB intensity values.
The main objective of this project in designing fuzzy logic is the determination of membership function in order to classify and to grade mango fruits.From the membership function, the range of size for mango fruits was obtained.In this study, an inductive reasoning method in designing fuzzy membership functions was used.This approach is based on minimizing entropy function which is partitioning a threshold line between three variables of data: length, width and height of mango fruit.The results show membership functions obtained are shaped from the threshold.On the other hand, this project also applied the IF-THEN Rules and Mamdani Implication to investigate the relationships which exist between the variables.

Figure
Fig Figur Figure 10