KLASIFIKASI CITRA DAUN TEMBAKAU VORSTENLANDEN BERDASAR CIRI WARNA (RGB) MENGGUNAKAN METODE KNEAREST NEIGHBOR; VORSTENLANDEN TOBACCO LEAF IMAGE CLASSIFICATION BASED ON COLOR FEATURE (RGB) USING K-NEAREST NEIGHBOR METHOD

Andaru, Nindya Putri, Agus Harjoko

2015 | Skripsi | FMIPA

Vorstenlanden tobacco sale and purchase agreement between farmers and the company as a buyer could not be separated from the process of quality assessment. There are several characteristics that are owned by the Vorstenlanden tobacco leaves to determine its quality, one of which is color. Determination of the quality of Vorstenlanden tobacco leaves which are done visually by a grader manually is perceived less thorough and less objective for significant number of leaves. Meanwhile, the development of information technology enables the determination of the quality to be more efficient. The program designed in this study is the classification program of Vorstenlanden tobacco leaf color, which is one of the parameters determining the quality of tobacco. This program will classify the image of tobacco leaf into three color classes, namely blue, yellow, and red. The method used is the Otsu segmentation and edge detection Sobel operator for segmentation stage, as well as k-Nearest Neighbor algorithm for the classification process. This research resulted success percentage of 43.33% for the Otsu method, and 72.67% for edge detection method.

Kata Kunci : Digital image processing; tobacco leaf; segmentation; classification