Abstract: Currently, controlling or identifying of the sugarcane white leaf disease infection is not possible since its symptom need to be observed by human walking throughout fields. Therefore, this research aims to study the ability of vegetation indices to detect white leaf disease infected sugarcane with images taken by multispectral camera mounted on Unmanned Aerial Vehicle. Three sub-images infected with white leaf symptom and 3 other sub-images of normal green leaf were selected for this study. The reflectance values of 6 chosen sub-images were used to calculate 18 vegetation indices, and then these indices were used to compute the difference percentage of vegetation (green versus white) in order to find vegetation indices that are the most sensitive to white leaf symptoms. The results show that vegetation indices that have NIR and Red edge band in their formula (14 vegetation indices) have difference percentage in the range of 14.66 – 45.10, with NDREI, GNDVI yielding the highest difference percentage (44.05 – 45.10%), and vegetation indices that have only visible bands in their formula (4 vegetation indices) have the difference percentage from 14.96 – 26.04%, with GI and NRI resulting in the highest difference percentage 24.04% and 26.04%, respectively.
Authors: P Sanseechan, K Saengprachathanarug, J Posom, S Wongpichet, C Chea and M Wongphati
Associations: Khon Kaen University, HG Robotics Company Limited, and University of the Ryukyus