Implementasi Ekstraksi Ciri Histogram dan K-Nearest Neighbor untuk Klasifikasi Jenis Tanah di Kota Banjar, Jawa Barat

Abstract

Land plays an essential role in the availability of nutrients and water to support our life on earth. Soil quality can be seen from the characteristics of color and texture. By knowing the quality of the soil, we can determine the type of plant that is most suitable for planting. In this study, we conducted a study of soil quality at Langensari. Langensari was chosen because most of the region has an altitude of fewer than 25 meters above sea level, so it is very potent as an agricultural and planta-tion area. The proposed system uses a cross-sectional image of the ground as input. The image is then extracted using histo-gram feature extraction to obtain the intensity, standard deviation, skewness, energy, entropy and smoothness values. K-Nearest Neighbor then used to classify resulting features. The proposed system was tested using 20 test images. Based on the experiment result, the system can classify soil types appropriately with accuracy reaching 60% when value of K = 3.