Wheat Disease Discrimination Utilising Wavelet-Based Models in the NIR Spectrum
||25 June 2019
||9:00 AM - 11:00 AM
||Toowoomba - T357, or via Zoom
||For more information, please contact the Graduate Research School.
|Save to calendar:
The near-infrared spectrum is investigated for the ability to discriminate wheat crown rot (Fusarium pseudograminearum) outside of the visible spectrum. Specifically, the dyadic discrete wavelet transform is utilised for the preprocessing of hyperspectral sensor data, to allow for the creation of crop disease models with wider bandwidths. The unique scaling ability of the dyadic discrete wavelet transform is applied to hyperspectral data obtained from crown rot resistance trials. Data features are extracted using supervised machine learning techniques and analysed using k-fold cross-validation and F-scores. The results show promise for automated classiﬁcation of disease with poor visible markers, using wavelet-derived models. The wavelet-based classiﬁcation system provides 95% classiﬁcation accuracy with as few as two regressors (i.e. wavebands).