Estimation of plant nitrogen content using digital image processing

Abstract

A manually operated four wheel test trolley was designed and developed for acquiring outdoor color images of plant under controlled illumination to predict crop nitrogen content in field.  This set up consists of a camera to capture the plant image, four lights to control illumination and a laptop for processing the signal.  The developed unit was evaluated rigorously for paddy crop for four observations at fifteen days interval after transplantation.  The results were compared with the chlorophyll content of the crop measured by SPAD meter and the chemical analysis of plant leaf.  The processing of the color plant image was done in MATLAB 7.0 program.  Various features such as R, G, B, normalized ‘r’ and normalized ‘g’ were analyzed for both the processes.  Regression models were developed and evaluated between various image feature and the plant nitrogen content and observed that, the minimum accuracy was found to be 65% with an average accuracy of 75% (Standard Deviation +1.9), actual and predicted values of nitrogen percent were linearly correlated with R² value (0.948), this showed that the plant nitrogen content can be successfully estimated by its color image feature.

Keywords: precision agriculture, digital image processing, site specific nitrogen application