Identifikasi Freycinetia Berbasis Citra Anatomi Epidermis Daun Menggunakan k- Nearest Neighbor
Abstract
Freycinetia identification was conducted to determine the potential values, benefits, and distribution patterns as a biodiversity in Indonesia. The purpose of this research was to compare the low frequency and high frequency in Freycinetia identification based on leaf epidermis anatomy image by k-Nearest Neighbor (k-NN). Identification of Freycinetia can be described by using morphology and anatomy characteristics. Leaf epidermis anatomy image was used in the identification process to support the morphology characteristics, especially in speciment and sample with incomplete morphology. This research analyzed ninety six data which contains four kinds of Freycinetia, namely Freycinetia angustifolia, Freycinetia imbricata, Freycinetia javanica, and Freycinetia Sumatrana. The data were transformed by Fourier transformation and filtered in frequency domain to take the low frequency of image and the high frequency of image. It was found that the accuracy of k-NN with low frequency was 90.625% and that of the high frequency was 81.25%. These accuracy values indicated that in the identification of Freycinetia based on leaf epidermis anatomy image by k-Nearest Neighbor (k-NN), the use of low frequency is better than high frequency
Collections
- UT - Computer Science [2322]