Search

Absorption and scattering properties of product change as moisture content is reduced, but it has not been investigated how these changes are correlated. This study was aimed to measure and test the feasibility of using optical properties in predicting the moisture content of sliced apple samples during hot air drying. In this investigation, the noninvasive backscattering laser imaging technique at three wavelengths (650, 780, and 880 nm) and Farrell’s diffusion theory were used to determine absorption and reduced scattering coefficients. Artificial neural network model was applied to correlate the optical coefficients and moisture content of samples. The highest correlation between above-mentioned parameters was found at 780 nm. The best moisture content prediction result was obtained when absorption and reduced scattering coefficients were combined at three wavelengths withRp = 0.984. The results suggested that this method can be effectively used to predict the moisture content and control the drying process. [ABSTRACT FROM PUBLISHER]

Highlights • Wavelet transform was used as pre-processing tool in Vis/NIR spectral analysis. • Developed models showed better regression performance than previous studies. • Wavelet transform showed acceptable performance for denoising of spectral data. Abstract Quality assessment of fruits and vegetables is important for all involved in production, processing, trading and even consumption of the products. This study was conducted for non-destructive estimation of moisture content (MC), soluble solids content (SSC), pH and firmness of Gala apple samples using near-infrared spectroscopy in the 350–2500 nm range by application of wavelet transform for pre-processing of raw spectral data. Wavelet transform was combined with other usual pre-processing functions widely used for constructing PLS regression models. The models' predictive capability was evaluated using correlation coefficient of determination (R2), root mean square error of estimation (RMSE) and correlation coefficient (R). The best calibration and validation results were obtained for MC with 8 factors, R2 C = 0.90, RMSEC = 0.0042, R2 CV = 0.88, and RMSECV = 0.0047. The PLS regression parameters for SSC with 9 model factors, were R2 C = 0.90, RMSEC = 0.37, R2 CV = 0.86, and RMSECV = 0.45. The calibration and validation model parameters for pH and firmness with 10 model factors were R2 C = 0.87, RMSEC = 0.04, R2 CV = 0.84, RMSECV = 0.05, and R2 C = 0.76, RMSEC = 2.23, R2 CV = 0.0.59, and RMSECV = 3.04, respectively. In the prediction set, the model that was developed for moisture content had RMSEP = 0.009, R2p = 0.6. The R2p, and RMSEP for SSC were, 0.87 and 0.55, respectively. The model prediction parameters for pH were 0.72 and 0.06 for, R2p and RMSEP, respectively. Finally, predicted firmness of apple samples were acceptable (R2p = 0.65 and RMSEP = 3.86). Because of noise removal ability, application of wavelet transform for pre-processing on spectral data led to accurate, simple and fast development of PLS regression models. [ABSTRACT FROM AUTHOR]

• A new approach for in-situ monitoring of apple nutrient contents was proposed. • Apple leaf spectra can replace costly and destructive analysis to predict NPK. • The RF model with Rfrog wavelength selection achieved the best results. • PLSR-based models are not recommended for nutrient estimation. • The approach can be used to develop a portable instrument to estimate NPK concentrations. Timely and rapid monitoring of apple trees nutrition status is vital for accurate management of nutrient fertilizers in order to improve the yield and quality, as well as to reduce the risk of environmental degradation. As a most frequently used method, tissue analysis used to assess the nutritional status of apple tree leaves is laborious, costly, time-consuming, environmentally unfriendly, and destructive. Ground-based sensors are able to efficiently provide information on nutritional status using leaf spectra reflectance. This research aims to establish a novel cost-effective and non-destructive approach for rapidly estimating the status of nitrogen (N), phosphorus (P), and potassium (K) in apple tree leaves based on Visible/Near-infrared (Vis/NIR) spectroscopy (500–1000 nm) coupled with machine learning. The Vis/NIR spectra of apple trees' leaves were acquired. Then, leaf chemical contents of NPK elements were considered as reference points. Different pre-processing techniques were used to pre-treat the spectra. Four different chemometrics analysis consist of support vector machine (SVM), Artificial neural network (ANN), Random Forest (RF) and partial least square (PLS) were applied to predict NPK contents in comparison to actual values. In order to simplify the models, the sensitive wavelengths were extracted using three wavelength selection approaches, variable importance in projection (VIP), partial least squares (PLS), and random frog (Rfrog). The extracted feature wavelengths from PLS, VIP, and Rfrog methods were widely distributed in the visible and near-infrared regions of the spectrum. The performance of the developed models was tested using Residual Prediction Deviation (RPD) and the Ratio of Performance to Interquartile Distance (RPIQ). The results demonstrated that among all models, the non-linear modeling methods were superior to the linear model. The best results for estimation of Nitrogen (N), Phosphorus (P), and Potassium (K) elements were achieved by the models of MSC + D 2 -Rfrog-RF (r p = 0.985, RMSEP = 0.029%, RPD = 8.77, RPIQ = 7.72), SNV + D 2 -Rfrog-RF (r p = 0.977, RMSEP = 0.0053%, RPD = 6.42, RPIQ = 5.09) and SNV + D 2 -Rfrog-RF (r p = 0.978, RMSEP = 0.018%, RPD = 8.16, RPIQ = 7.01), respectively. The findings of the current approach may provide an efficient approach to predict in-situ NPK contents of apple trees based on leaf spectral reflectance. [ABSTRACT FROM AUTHOR]