研究目的
1) to evaluate the potential of close-range imaging spectroscopy data for detecting rice leaf blast disease and 2) to examine the spatial and temporal variation in leaf biochemistry in response to the disease with a physical inversion method.
研究成果
Preliminary results demonstrated that the retrieved chlorophyll maps exhibited spatial heterogeneity within the leaf and reasonable LCC values for healthy and infected portions of a leaf. Anomaly regions corresponding to disease infestation could be clearly seen from PROCWT-retrieved LCC maps and even larger than the visible symptoms on the true color images. The LCC maps from model inversion provide an opportunity to analyze the temporal evolution and the spatial variation of the symptoms from the pre-visual stage to the visual stage.
研究不足
The study only reported on the use of LCC maps for examining the anomaly regions related to rice leaf blast. Future work is directed to the retrieval of other biochemical variables including carotenoid and anthocyanin contents for a larger number of samples. Water content may also be a valuable variable to examine but its retrieval requires the spectral information from the shortwave infrared (SWIR) region.
1：Experimental Design and Method Selection:
A greenhouse experiment with artificial inoculation of rice blast fungal was conducted. Hyperspectral images were acquired from diseased leaves at different infection stages. The image data were converted to reflectance cubes and processed with a model inversion algorithm PROCWT to retrieve leaf biochemical variables.
2：Sample Selection and Data Sources:
Three locally commercial varieties of Japonica rice were planted in a total of 36 buckets, with 18 for artificial inoculation and 18 for health protection.
3：List of Experimental Equipment and Materials:
A push-broom hyperspectral imaging system consisting of a computer, a hyperspectral camera (GaiaField-F-V10E camera), two halogen lamps, a 3% reflectance blackboard and a black box.
4：Experimental Procedures and Operational Workflow:
The images were acquired at near-nadir viewing position at a spatial resolution of about
5：22 mm. Three fully developed leaves were chosen and fixed on the 3% reflectance blackboard to minimize the background impact. Data Analysis Methods:
The leaf chlorophyll content (LCC) of rice samples was retrieved with two methods: the physical model inversion approach PROCWT and the empirical model approach WREP.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
