Research and analysis of cadmium residue in tomato leaves based on WT-LSSVR and Vis-NIR hyperspectral imaging

The reliability and validity of Vis-NIR hyperspectral imaging were investigated for the determination of heavy metal content in tomato leaves under different cadmium stress. Besides, a method involving wavelet transform and least square support vector machine regression (WT-LSSVR) is proposed to select the optimal wavelength and establish the detection model. The Vis-NIR hyperspectral images of 405 tomato leaf samples were obtained and the whole region of tomato leaf sample spectral data was collected and preprocessed. In addition, WT-LSSVR is used to select optimal wavelength and establish the detection model using db4 and db6 as wavelet basis function, respectively. Furthermore, the best prediction performances for detecting cadmium (Cd) content in tomato leaves was obtained by second derivative (2nd Der) pre-processing method, with R_c² of 0.9437, RMSEC of 0.0988 mg/kg, R_p² of 0.8937, RMSEP of 0.2331 mg/kg, R_cv² of 0.9357, RMSECV of 0.1455 mg/kg, RPD of 3.081 and bias of 0.00863 using db6 (daubechies 6) as wavelet basis function with wavelet fourth layer decomposition. The results of this study indicated that WT-LSSVR can effectively select the optimal wavelength and Vis-NIR hyperspectral imaging has great potential for detecting heavy metal content in tomato leaves under different cadmium stresses.