Early Disease Detection of Bacterial Leaf Blight on Rice Plant by using Hyperspectral Imaging = Early Disease Detection of Bacterial Leaf Blight on Rice Plant by using Hyperspectral Imaging

다국어 초록 (Multilingual Abstract)

In Southeast Asia, bacterial leaf blight (BLB) is one of the significant rice diseases that potentially causes productivity loss up to 50%. The early stage disease detection contributes to prevent the serious crop damage due to the disease outbreak. The hyperspectral imaging provides an extension of our vision by capturing the near infrared (NIR) region at 700-1000 nm which is the outside of visible light (VIS) region at 400-700 nm. In this study, the possibility of early stage disease detection method of BLB on rice plant was investigated in captured hyperspectral images with 40-50 cm from the top of rice plant. For early disease setting, the hyperspectral images for both the healthy and infected single leaf of the rice plants, both were taken from 2 days post-inoculation when the symptom had not been shown yet. The possible sensitive bands were explored from VIS to NIR and chosen by applying the principal component analysis (PCA) on reflectance value of the healthy and infected leaves using 400-850 nm. The PCA result provided that there several ranges of wavelength have high correlation with the disease development on early stage and found that the ratio of orange-red color (605-640 nm) and red edge (690-720 nm) has the highest possibility to detect the BLB.