Utilizing Spectral Indices on UAV Multispectral Images for Paddy Healthiness Detection: A Case Study in Perlis, Malaysia

Abstract

The increasing global population has brought challenges in expanding and maintaining the productivity levels of paddy. Nowadays, the use of Unmanned Aerial Vehicles (UAV) and multispectral sensors in precision farming has become a prevalent approach in the agriculture sector to enhance efficiency, production, and sustainability in various agricultural activities, including paddy cultivation. In addition, the red edge spectral in multispectral sensor which reflects the rapid change in vegetation is the most suitable for crop studies and very significant to be applied in the computation of spectral indices. Thus, the study aims to utilize various spectral indices on UAV Multispectral Images for the detection of paddy healthiness levels. Six (6) significant Vis (Vegetation Index) i.e., Normalized Difference Red Edge Index (NDREI), Normalized Difference Vegetation Index (NDVI), Optimized Soil Adjusted Vegetation Index (OSAVI), Soil Adjusted Vegetation Index (SAVI), Nitrogen Reflectance Index (NRI) and Green Normalized Different Vegetation Index (GNDVI) were computed and analyzed to determine the affected and healthy paddy of study areas. It was found that the NDREI gave the best accuracy in classification and significant results compared to other indices. These could be due to the application of the Red-Edge band in the algorithm used by NDREI. Meanwhile, the NRI has the lowest accuracy in classifying the paddy area due to its insensitivity to infected paddy. Overall, the severeness of infected and healthy paddy plants can be detected from the computation spectral indices on UAV multispectral, particularly with the red edge spectral band which can provide a comprehensive paddy healthiness levels in the area.