Today, keeping and determining the quality of the food products is actually a problem. The quality control workers classify the fruits and vegetables according to their quality and properties by means of hands and eyes. However, this method does not provide the standard of quality and it is possible to occur incorrect classification. At the same time, there is a great loss in terms of time and labour force. In order to eliminate these cases, it is possible to categorize fruits quickly and in line with the standards depending upon a machine automatically. In this study, it is aimed to ascertain the dimension, colour, category, weight of the apples moving on a band in a real time. With a color camera situated on the closed cabin and on a moving band, it is possible to take and process the image of apples in real time: The weight and the size of the apples are estimated by using their areas on the acquired image before the classification. The category information of the apples can be determined by color analysis. Moreover, the defects and the symptoms of illness can be determined and new separation can be processed if required. The developed MATLAB program allows users to acquire the images of apples, change the speed of conveyer band and classify the apples. The apple images, category stamps and the data on category statistics can be seen on the screen. This program keeps the user information and also the data of daily, weekly, monthly and yearly production in a database. In the database, the eight kinds of apple cultivated in Isparta region, their categories and color are stored. The developed system can classify an apples in a half second time on average with the success rates of 95.52% and 99.00% for category and color estimation, respectively. Better rates can be obtained for weight and dimension estimation. As results, fast and high accurate apple classification can be carried out by the developed system. Despite the high cost at initial settings, the computer based automatic apple classification system reduces the processing cost in long turn. Keywords: Image Processing, Apple Classification, Fruit Classification
Tez (Yüksek Lisans) - Süleyman Demirel Üniversitesi, Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Anabilim Dalı, 2011.
Kaynakça var.
Today, keeping and determining the quality of the food products is actually a problem. The quality control workers classify the fruits and vegetables according to their quality and properties by means of hands and eyes. However, this method does not provide the standard of quality and it is possible to occur incorrect classification. At the same time, there is a great loss in terms of time and labour force. In order to eliminate these cases, it is possible to categorize fruits quickly and in line with the standards depending upon a machine automatically. In this study, it is aimed to ascertain the dimension, colour, category, weight of the apples moving on a band in a real time. With a color camera situated on the closed cabin and on a moving band, it is possible to take and process the image of apples in real time: The weight and the size of the apples are estimated by using their areas on the acquired image before the classification. The category information of the apples can be determined by color analysis. Moreover, the defects and the symptoms of illness can be determined and new separation can be processed if required. The developed MATLAB program allows users to acquire the images of apples, change the speed of conveyer band and classify the apples. The apple images, category stamps and the data on category statistics can be seen on the screen. This program keeps the user information and also the data of daily, weekly, monthly and yearly production in a database. In the database, the eight kinds of apple cultivated in Isparta region, their categories and color are stored. The developed system can classify an apples in a half second time on average with the success rates of 95.52% and 99.00% for category and color estimation, respectively. Better rates can be obtained for weight and dimension estimation. As results, fast and high accurate apple classification can be carried out by the developed system. Despite the high cost at initial settings, the computer based automatic apple classification system reduces the processing cost in long turn. Keywords: Image Processing, Apple Classification, Fruit Classification