A comparison of visual and digital image-processing methods in quantifying the severity of coffee leaf rust (Hemileia vastatrix)

Abstract

Using disease diagrams, visual assessments of coffee leaf rust from Papua New Guinea overestimated severity by >2 times when >20% leaf area was rusted. Nevertheless, a significant linear relationship was established between visual and planimeter assessments of coffee rust severity. Colour image analysis, using a Matrox digitiser, was more accurate than visual estimates or black and white imaging systems in discriminating between rusted and healthy leaf tissues in colour transparency photographs of rusted coffee leaves. Electronic flash photography of rusted coffee leaves produced images with reflections from the waxy leaf surface which made it difficult to discriminate between healthy tissues and the background.