K-Nearest neighbors (k-NN) algorithm to classify nitrogen nutrient rice plants status based on leaf color chart (LCC)

Muliady Muliady, Lim Tien Sze, Koo Voon Chet, Suhadra Patra


Soil nutrients had the biggest impact on the growth and quality of rice plants, especially Nitrogen. Rice that grew on Nitrogen-deficient soils would have the leaves’ color yellowish-green. The gradation of yellow-green color showed the level of Nitrogen deficiency. Comparing Leaf Color Chart with the color of rice leaves would give the Nitrogen deficiency status, but human eyes were affected by the surrounding lighting. This will lead to inappropriate estimation of the level of Nitrogen deficiency. One of the solutions is implementing a machine-learning algorithm to help to classify the leaves' color. The references showed some issues still exist. This research proposes the idea to create the training data and testing data with consistency lighting and add the light intensity information to the classifier. The rice leaves training data were taken in the morning sunlight at 6.00-7.00 (UTC+7:00) with 2000-3500lux. A total of 84 training data were created from 21 images for each LCC value. A smartphone application was built to measure and prevent the user take an image if the light intensity is not in 2000-3500 lux. By using cross-validation method founded k=5 gave the best result with the success rate of k-NN classifier at 97,22%. 


Consistency lighting; k-NN; Leaf color chart; Light intensity; Nitrogen nutrient; Rice plants

DOI: http://doi.org/10.11591/ijeecs.v22.i1.pp%25p
Total views : 30 times


  • There are currently no refbacks.

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

shopify stats IJEECS visitor statistics