Identifikasi Daun Shorea Menggunakan K-Nearest Neighbor dengan Ciri Statistical Textures

Abstract

Various species of Shorea from Dipterocarpaceae family are dominating the lowland rain forests in Indonesia. Each species of Shorea has a unique function and can produce a high qualified timber in the construction industry. This research built an identification system of Shorea leaves using the statistical textures feature extraction with k-nearest neighbor as classifier. Statistical methods were utilized to analyze the spatial distribution of gray level values, by computing local features at each point in the image and deriving a set of statistical values from the distribution of the local features. The calculation of gray level values resulted in six parameters, namely mean, standard deviation, smoothness, third moment, uniformity, and entropy. A highest accuracy of 90% was obtained for the identification of 10 species of Shorea