Author Affiliations
Institute of Optics Mechanics Electronics Technology and Application, East China Jiaotong University, Nanchang, Jiangxi 330013, Chinashow less
Fig. 1. Lossless, regular-shaped blade
Fig. 2. Capture schematic of chlorophyll fluorescence spectrum
Fig. 3. Original full spectrum image
Fig. 4. Fluorescence spectrum pretreatment and model comparison process
Fig. 5. Fluorescence spectra of fresh tea leaves after smoothing treatment
Fig. 6. Variable stability results of UVE method screening
Fig. 7. Results of variable selection using UVE method
Fig. 8. Results of variable selection using SPA algorithm. (a) Relationship between number of selected variables and RMSEP; (b) locations of selected variables in processed spectra
Fig. 9. Prediction results of simplified PLS model
Sample | Maximum | Minimum | Mean | SD | CV |
---|
Tea leaf | 87.0±0.3 | 61.6±0.3 | 78.93 | 5.09 | 0.06 |
|
Table 1. Statistics on chlorophyll content
Evaluation index | Before | After |
---|
3 points | 5 points |
---|
RC | 0.8339 | 0.9129 | 0.9162 | RMSEC | 2.2360 | 2.2776 | 0.2747 | RP | 0.8820 | 0.8675 | 0.8989 | RMSEP | 1.8811 | 1.9199 | 1.4831 |
|
Table 2. PLS values before and after chlorophyll relative content treatment
Subset | Number of samples | Minimum | Maximum | Mean | SD |
---|
Calibration set | 80 | 61.6±0.3 | 87.0±0.3 | 78.56 | 5.80 | Prediction set | 40 | 70.9±0.3 | 85.8±0.3 | 79.66 | 3.16 |
|
Table 3. Descriptive statistics for calibration set and prediction set
Method | Number of variable points | Calibration set | Prediction set |
---|
RC | RMSE | RP | RMSE |
---|
PLS | 1044 | 0.9063 | 2.4359 | 0.9386 | 1.4871 | UVE-PLS | 160 | 0.9174 | 2.2208 | 0.9224 | 1.4073 | SPA-PLS | 10 | 0.9176 | 2.2896 | 0.9450 | 1.3069 | BiPLS | 280 | 0.9048 | 2.4545 | 0.9392 | 1.4861 | BiPLS-SPA | 14 | 0.9612 | 0.9483 | 0.9591 | 0.8863 |
|
Table 4. Descriptive statistics of calibration set and prediction set
Numberintervalsin model | Selectedinterval | Numberofvariables | RMSECV | Numberintervalsin model | Selectedinterval | Numberofvariables | RMSECV |
---|
20 | 6 | 400 | 2.6825 | 10 | 11 | 200 | 2.3902 | 19 | 9 | 380 | 2.4252 | 9 | 5 | 180 | 2.3643 | 18 | 4 | 360 | 2.3616 | 8 | 14 | 160 | 2.3413 | 17 | 18 | 340 | 2.3325 | 7 | 3 | 140 | 2.3402 | 16 | 19 | 320 | 2.3269 | 6 | 1 | 120 | 2.4077 | 15 | 20 | 300 | 2.3222 | 5 | 2 | 100 | 2.4248 | 14 | 17 | 280 | 2.3212 | 4 | 10 | 80 | 2.3955 | 13 | 16 | 260 | 2.3221 | 3 | 13 | 60 | 2.7288 | 12 | 15 | 240 | 2.3333 | 2 | 12 | 40 | 2.7245 | 11 | 8 | 220 | 2.3684 | 1 | 7 | 20 | 2.4189 |
|
Table 5. Results of selecting optimal intervals step by step