NPK Soil Nutrient Measurement Prototype Based on Local Binary Pattern And Back-Propagation

Abstract

Nutrient elements of NPK are macro nutrients that play an important role in the growth and development of plants, therefore it is necessary to measure NPK nutrient content to measure how well soil fertility condition before the land planting period, but NPK measurement through laboratory tests takes a relatively long time. This research develops a prototype of NPK nutrient measurement system based on a mobile application by using soil image for determining the textural characteristic, the textural characteristics are processed with local binary pattern and back-propagation neural network to accelerate the measurement process.Sample data in this research was taken on rice field land in the province of Yogyakarta Special Region by varying the distance at 30 cm to 110 cm with interval 20 cm and angle image capture at −30° to 30° with interval 10°. Datasets were being pre-processed to improve image quality and adjust image format. Preprocessed results are extracted using local binary pattern uniform to obtain texture features. The texture features were being inputted of the neural network model, that being trained with a back-propagation algorithm by varying parameters of the neural network model.The model tested to determine the effect of distance and angle of image capture, system processing speed, and effect of artificial neural network parameters. The best model is implemented on a smartphone application. The results obtained an average of computation time 0.65s, and the optimal result is obtained at distance capture of 50 cm and angle capture of 0° with the measurement accuracy at each soil nutrient level of nitrogen 91.80%, while phosphorus 83.49%, and potassium 82.54%, therefore the average is 84.16%