Nondestructive detection of rape leaf chlorophyll level based on Vis-NIR spectroscopy.

Chlorophyll is an important factor for measuring the normal growth and development status of plants, and it is also of great significance for the management and utilization of agricultural water and fertilizers. In this study, the chlorophyll content of rapeseed leaves was taken as the research object, and the effect of spectral data pretreatment method on the spectral feature extraction and chlorophyll content prediction model was quantitatively studied. ASD FieldSpec Pro (350-2500 nm) spectrometer was used to measure the spectral reflectance of rape leaf samples, and the spectral reflectance characteristics of different chlorophyll contents were analyzed. The Savitzky-Golay nine-point smoothing of the reflectance spectrum was performed, and the first derivation (FD), second derivation (SD), and reciprocal logarithm (LOG)transformation of the reflectance were performed after MSC and SNV preprocessing respectively. The optimal spectral estimation model for chlorophyll was established by PLSR. The results show that: (1) This study was mainly to monitor the chlorophyll content in rape joints during jointing stage, using the correlation between chlorophyll content and hyperspectral characteristics, using MSC, NOR and SNV to pretreat the reflectance spectra and combining different derivations transformations to extract chlorophyll characteristics. (2) Quantitative model of chlorophyll content was established based PLSR, the best preprocessing was R + SG + SNV + LOG+FD, the calibration results was: LVs = 14, Rc2 = 0.97, RMSEC = 4.18, SEC = 4.21, Slope = 0.92, Offset = 2.63; the validation results was: Rv2 = 0.98, RPD = 7.52, RMSEP = 2.94, SEP = 2.98, Slope = 0.98, Offset = 1.43; (3) The optimal estimation model established by different treatment methods has better stability and higher precision, and can rapidly monitor the chlorophyll content of rapeseed in the region.