Detection of Fusarium Head Blight of Wheat from hyperspectral images

The increasing occurrence of stresses in agricultural crops due to pathogens causes a general decline in productivity, becoming a major risk factor for safety and food security. Fusarium Head Blight is a major plant disease characterized by specific pathogenic mechanisms and symptoms, such as spike bleaching, and it represents one of the major biotic threats affecting Mediterranean countries. Remote sensing and field scale optical techniques represent an appealing approach for improving crops monitoring and early detection of diseases. A small-scale controlled field experiment was organized to assess the response of wheat (Triticum aestivum L.) cultivars to Fusarium infections. Hyperspectral images of wheat spikes were acquired in laboratory with a novel visible to near-infrared hyperspectral imaging system with the intent to develop methods capable to automatically identify Fusarium-induced symptoms. We tested the capability of the so-called phasor-based approach to highlight spikes areas affected by Fusarium infection. Outcomes show that infected spikes do not lie completely superimposed to the healthy points cloud in the phasor plane, suggesting that (i) infected grains are characterized by different spectral features, and (ii) Fusarium-related symptoms are not extended throughout the entire spike. This contribution shows the methodology developed and the results obtained from the analysis of hyperspectral images.