SISTEM KLASIFIKASI KEMATANGAN TOMAT BERDASARKAN WARNA DAN BENTUK MENGGUNAKAN METODE SUPPORT VECTOR MACHINE (SVM)

Abstract

Ordinary tomatoes (Lycopersicun Commune) fruit shape round flat and size is not regular. This type of tomato is very suitable to be grown in lowland areas. This shows that tomato commodity has been consumed by society widely and have competitiveness. During this sorting process is still done manually that still has many shortcomings. Manual classification gives a classification result that is less precise and inconsistent due to the negligence of humans. Quality in the processing and marketing sectors is important. Improper quality has the potential to harm the farmers because all the quality of fruit is equal. For that we need a consistent classification system. The system uses image processing to extract color and shape features. The classification method used is the Support Vector Machine (SVM). This system will classify the tomatoes into 3 quality classes, namely class A, class B, and outside of quality I, beyond quality II. SVM is designed with an input of 10 extraction features ie average RGB value (Red, Green, Blue), and AP value (area and perimeter) with 3 class output. Digital image processing used for tomato objects from digital cameras will generate the intensity of reflections that illuminate the light and dark on the appearance of the pixel-pixel arrangement and will also provide color information, pixels can be shown its location by using coordinates. Testing using 10 x 10 fold crossvalidation method, from the test results obtained that the system is able to provide an average accuracy of 82.83% and standard deviation 1.52.