Tutorial II: Disease detection with fusion techniques

Abstract In the current case study, the fusion of features derived from multispectral sensors and hyperspectral cameras was employed to discriminate induct from infected wheat aiming to further utilize it as a detector for crop biotic stress. Spectral reflection has been employed in order to discriminate water stress from symptoms of Septoria appearance by the use of a hyperspectral camera. At the same time, the crop health status was evaluated through fluorescence kinetics measurements with the help of a Plant Efficiency Analyser (PEA) fluorimeter (Hansatech Instruments, Norfolk, UK). A data fusion approach was combined with LSSVM approach aiming to discriminate water stress from infected wheat plants. The LSSVM is evaluated against various classifiers including the MLP and QDA for detecting and separating water stress plants from the diseased ones.