RGB vegetation indices applied to grass monitoring: a qualitative analysis

Abstract

In developing countries such as Brazil, research on low-cost remote sensing and computational techniques become essential for the development of precision agriculture (PA), and improving the quality of the agricultural products. Faced with the scenario of increasing production of emerald grass (Zoysia Japônica) in Brazil, and the value added the quality of this agricultural product. The objective of this work was to evaluate the performance of RGB (IV) vegetation indices in the identification of exposed soil and vegetation. The study was developed in an irrigated area of 58 ha cultivated with emerald grass at Bom Sucesso, Minas Gerais, Brazil. The images were obtained by a RGB digital camera coupled to an remotely piloted aircraft. The flight plan was setup to take overlapping images of 70% and the aircraft speed was 10 m s -1 . Six RGB Vegetation index (MGVRI, GLI, RGBVI, MPRI, VEG, ExG) were evaluated in a mosaic resulting from the images of the study area. All of the VIs evaluated were affected by the variability of lighting conditions in the area but MPRI and MGVRI were the ones that presented the best results in a qualitative evaluation regarding the discrimination of vegetation and soil.