Aplikasi Digital Image Processing untuk Grading Citra Manggis

Abstract

Mangosteen (Garcinia Mangostana Linn) is an Indonesian native fruit with the nickname of "queen of fruit". Indonesia has a
very high export potential for mangosteen. Banyuwangi is the largest production center of mangosteen fruit in East Java. The
main problem often faced by mangosteen farmers in the post-selection process is the quality selection which is done manually. It
raises several issues, such as less equitable observations due to visual limitations of fatigue and observers’ differences in
perceptions. Depend on these problems, taking pictures with camera and digital image processing could be a solution. Digital
image processing was done by extracting the color of the mangosteen with various quality classes into Hue, Saturation and
Intensity and by measuring the diameter to be used astesting data. The qualification process of mangosteen fruit used k-NN
method in which the sorting was done based on the proximity of the distance between the image features and the diameter which
were obtained from the tested mangosteen toward the training data. In this case study, the test was performed on 15 testing data
toward 75 training data consisting of various quality classes on the side view images. The test performed had the accuracy of
93.3% with the score of k = 3 and 86.6% with the score of k = 5 on the side view.

Keywords— Mangosteen, Grading, HSI, k-NN