Nitrogen Fertilizer Prediction of Maize Plant with TCS3200 Sensor Based on Digital Image Processing

Abstract

Fertilization for maize plants so far does not use the appropriate calculation of crop needs, thus it causes continuous soil damage. Another technique to predict nitrogen content in maize plants is using a leaf color chart. But it has some disadvantages, such as depending on sunlight for color plant reading, and inconsistent accuracy interpretation of leaf color by the eye. Based on problems, the research was aim to design a nitrogen content prediction system based on digital images with an RGB color index. RGB digital image and nitrogen content are used as a database system to predict nitrogen need for maize plants. Arduino UNO is used for data processor and the TCS3200 color sensor is used as a substitute for a camera on image processing. The performance test uses 40 leaves of maize plants with the age of 15, 30 and 40 days. The results show that the device can detect nitrogen content in maize plants based on the RGB database with a 90% success rate. For R value 447,4-601,6, G value 384,7-510, and B value 519,95-625,16, we obtain that nitrogen need for plant about 175 kg/ha. For R value 325,9-377,5, G value 285,6-311,5, and B value 426,4-456,1, we obtain that nitrogen need for plant about 150 kg/ha. And for R value 290,37, G value 239,12, and B value 385,62, the nitrogen need for plant about 125 kg/ha. The greener the color of maize leaf, the lower the nitrogen fertilizer need, conversely the lower the green color on the leaf, the higher the level of nitrogen fertilizer needs.