Abstract:
Grapevine Leafroll Disease (GLD) is a widely spread disease and major threat in grapevine (Vitis vinifera L.) cultivation. This causes reduction in yield and results poor quality fruits. Therefore, the GLD has become an intense area of research with great importance among the viticulture scientists and even with respect to ecology and management aspects. Several studies have been carried out using various techniques in order to find the appropriate remedial solutions for this virus infection. In order to control and minimize the impact of GLD, the early and accurate detections are essentially required. The use of optical sensing/imaging techniques for exploring biological activities has become very effective and hence popular due to promising results with high accuracy where the unseen world could be seen with more clear visuals. Several advanced optical sensing techniques are being used for precision agricultural applications especially for plant disease. Spectroradiometry is one of the fastest and accurate techniques based on light radiation, which enables to obtain the spectral reflectance patterns of the samples. The potential use of spectral reflectance techniques on detecting plant diseases have been verified by several studies. The accuracy of the method mainly depends on the spectral feature selection and analysis techniques. In our study, spectral data were collected from healthy and GLD infected grapevine leaves (both symptomatic and asymptomatic) on three different sampling days in the growth period of the grapevine. On each sampling day 60 spectra of healthy and infected leaves were obtained. Suitable feature extraction using stepwise linear regression and stepwise discriminant analysis reported significant visible differences between the healthy and infected plants (526, 626 nm) and near infrared (826, 901, 951, 976 nm) regions. The results from this study will be helpful in developing a low-cost field-portable sensing system for early detection of virul diseases of plant leaves.