Aplikasi Penentuan Dosis Kebutuhan Pupuk Nitrogen Berdasarkan BWD Pada Tanaman Padi

Abstract

Nitrogen is one of the nutrients that is needed for vegetative growth of rice, but excessive fertilizer application can damage plants. The balanced application of nitrogen fertilizer to rice plants is a solution to improve the growth of rice plants so that their productivity becomes more optimal. To get the appropriate fertilizer dose, farmers must use the BWD table, but BWD is difficult to obtain, the price is quite expensive, and its use is done manually by comparing the color of the rice leaves with the color of each level in the BWD table. Different perceptions between each use often occur. Therefore, an application for determining the need for nitrogen fertilizer based on BWD was designed for rice plants. This system consists of pre-processing, feature extraction and classification stages. The pre-processing stage is the stage of improving image quality, while the feature extraction stage uses the histogram of s-RGB method to obtain the Mean and Mode values ​​of the color intensity of rice leaves. This system classifies based on the characteristics that have been extracted into 3 classes, namely: 2-3, 3-4, and 4-5 based on the BWD level. Then the system will calculate the dose of nitrogen fertilizer needed based on the input data of GKG and land area. The classification stage uses the K-NN method. Based on the results of training using 210 images and testing 90 images of rice leaves, the best results were obtained using k-NN 3 neighbors with an accuracy of 95.5%, AUC 0.98 and training time 0.8 seconds. So it can be concluded that the classification using k-NN can determine the dose required for rice plants properly.