Estimating the crude protein contents in rice grains using hyperspectral data

The crude protein content in rice grains is one of the important indices to evaluate the nutrition and taste quality of rice. Normal determination of their contents by chemical methods are of high cost and time consuming. The hyperspectral reflectances of the canopy and flag leaf on main stem of two rice varieties were measured by a ASD FieldSpec Pro FR^TM in field under three nitrogen support levels in mature process. The nitrogrn contents of stems, leaves, panicles and cargo rice and their crude protein contents were determined. The correlation among them were analyzed. The crude protein contents of panicles and cargo rice were very significantly correlative to that of stems and leaves at booting, milking and maturing stages, and also very significantly correlative to the spectral reflectance ρ_λ, the first derivative spectra D_λ and RVI of canopy at some wave bands at milking and maturing stages. The crude protein contents of panicles were very significantly correlative to some hyperspectral indices of flag leaves on main stems. This indicated that it would be feasible to estimate the contents of nitrogen and crude protein in rice grains by hyperspectral remote sensing. It would provide basis for monitoring rice quality by remote sensing.