RGB Image-Based Coffee Rust Detection: Application of Vegetation Indices and Algorithm Development

Research and monitoring have focused on the growth and yield of agricultural crops, employing tools like vegetation indices (VIs) in the context of precision agriculture (PA). In this study, the implementation and performance evaluation of ten VIs were carried out to identify coffee leaf rust (CLR, Hemileia vastatrix). The indices evaluated were: Green Leaf Index (GLI), Excess of red (ExR), Excess of green (ExG), Excess of blue (ExB), Visible atmospherically resistant index (VARI), Normalized green-red difference index (NGRDI), Modified greenred vegetation index (MGRVI), Normalized pigment chlorophyll ratio index (NPCI), Triangular greenness index (TGI) and Color Index of Vegetation Extraction (CIVE). Subsequently, an algorithm was designed to identify CLR using the two VIs that showed the best results. Cohen’s kappa coefficient was used to assess inter-rater agreement. The VIs that exhibited the highest performance were NGRDI (77%) and MGRVI (81%). The algorithm developed based on these VIs achieved a 94.5% effectiveness in detecting CLR. The results obtained provide evidence of the efficacy of VIs in detecting rust, emphasizing their significance in developing novel algorithms specifically designed for CLR detection, mitigating the economic losses associated with it.

I. Introduction

The growth of the global population has resulted in an increased demand for food. Compounding this issue is the decrease in available cultivable land, necessitating the search for technological solutions that can complement agriculture and enhance productivity [1]. Precision agriculture (PA) emerges as a response to this challenge. PA is based on the utilization of autonomous sensors for close-range or remote detection, monitoring multiple factors relevant to crop health, including temperature, humidity, vegetation, among others. The main objective of PA is to provide farmers with precise parameters about the state of their crops, identifying areas that require attention and the necessary measures to be taken [2].