Identifying the spectral signatures of the broadleaved weeds and the rice plant through Hyperspectral Imaging Sensor (Resonon Pika L)
2023
Sulaiman, N. | Che'Ya, N.N. | Juraimi, A.S. | Noor, N.M. | Roslima, M.H.M.
Weeds create complications for rice (Oryza sativa L.) crops as they cause significant yield loss in yield production. The issue of broadleaved weeds in paddy fields poses significant challenges for rice farmers worldwide. Broadleaved weeds are undesirable plants that compete with rice crops for nutrients, water, and sunlight, ultimately reducing the overall yield and quality of the harvest. At the same time, it is difficult to differentiate them for site-specific weed management. Remote sensing technologies have emerged as powerful tools to capture and analyze the spectral characteristics of vegetation non-invasively. Identifying plant species based on their unique spectral signatures is crucial in various fields, including agriculture. This study presents a comprehensive analysis of identifying plant spectral signatures of the selected rice variety (MR297 and MR315) and the broadleaved weed species (Sphenoclea zeylanica, Monochoria vaginalis, Limnocharis flava) using remote sensing techniques through Hyperspectral Imaging Sensor (Resonon Pika L). Hyperspectral imaging captures a continuous spectrum of narrow and contiguous wavelength bands, resulting in a much finer spectral resolution. This detailed spectral information allows for the identification and discrimination of subtle differences in the reflectance properties of plants, including weeds and crops. Pre-processing steps ensure accurate and reliable data, including radiometric calibration, atmospheric correction, and geometric registration. In order to distinguish between plant species, advanced machine learning algorithms, including support vector machines, random forest trees, and neural networks, are employed to classify and differentiate spectral signatures, which provide more than 90% accuracy.
显示更多 [+] 显示较少 [-]