Vernonia amygdalina chlorophyll content prediction by feature texture analysis of leaf color

Abstract

Vernonia amygdalina has been scientifically proven to have activity against various diseases because it contains high antioxidants. The antioxidant content can be related to the chlorophyll content in leaves. Chlorophyll levels increase when the leaves are fully developed which is accompanied by an increase in antioxidants. So, chlorophyll detection by noninvasive sensing can be used to estimate the antioxidant content. An artificial neural network (ANN) was used to model RGB color as input and leaf chlorophyll content as output. Performance comparisons in each ANN model were carried out to find the best model in predicting leaf chlorophyll content, indicated by the smallest prediction error value. This study aims to model the chlorophyll content of Vernonia amygdalina with ANN analysis. The results showed that the chlorophyll content could be identified using 9 selected color texture features through the filter method feature selection with the best attribute of correlation. The selected ANN structure produces R training of 0.98522, R validation of 0.93417, MSE training of 0.0067, and MSE of validation of 0.0322. The results showed that digital image processing and ANN models have the potential as sensors in detecting the percentage of chlorophyll content of Vernonia amygdalina.