Online apple grading using image processing and artificial intelligence

Farhadi, Sonyia (2019) Online apple grading using image processing and artificial intelligence. Masters thesis, University of Mohaghegh Ardabili.

Text 1575098382318_سونیا فرهادی.pdf Download (469kB)

Abstract

Research Aim:Iran is the seventh largest apple producer in the world with 1.7 million tonnes of apples.Improper sorting results in the waste of a large part of the product and results in a waste of cost, time and labor. Today, the visual machine is one of the most basic methods used to identify and rate products. In this study, in addition to investigating the physical, mechanical and chemical properties of Golden Delicious apples,we investigated the relationship between these properties and the color parameters of the apples and also performed an algorithm for online detection of stained and crushed apples from healthy apples. Research method:Experiments and measure ofphysical, mechanical and chemical were carried out using a fruit sorting machine. Matlab software was used for writing the online image processing algorithm based on Apple Golden Delicious. Also part of the validation step of the algorithm was written using artificial neural network and part with MSTATC software and factorial test in a completely randomized design. Findings: The results showed that RGB color space conditions, moonlight and 0.2 m/s were desirabile for optimum online detection.Also the highest correlation coefficient was related to the red color of RGB images taken from apple sample with its pH value. Also found that the best neural network structure for predicting optimum imaging conditions is the use of 10 neurons in the hidden layer which achieved 0.972 correlation in the neural network test. Conclusion:Finally, the system was evaluated according to its intended purpose, which was able to accurately isolate 92.5% of healthy apples from stained apples.

Actions (login required)

View Item