Prediction of soluble solids content and anthocyanin content in blood oranges based on hyperspectral reflectance and transmittance imaging technologies

Anthocyanins and soluble solids content (SSC) serve as key factors for evaluating blood orange quality. Currently, reliable non-destructive measurement methods are lacking in production. In this study, hyperspectral diffuse reflectance and transmittance imaging (400 nm–1,000 nm) technologies were utilized to predict SSC and anthocyanin content in blood oranges. Three methods including standard normal variate (SNV) correction, moving average smoothing (MAS), and first derivative (Deriv1) were employed for preprocessing spectra. Additionally, bootstrapping soft shrinkage (BOSS), successive projections algorithm (SPA), and competitive adaptive reweighted sampling (CARS) were used to select effective wavelengths. Finally, partial least squares regression (PLSR) models were developed for predicting anthocyanin content and SSC in blood oranges. The results showed that the hyperspectral transmittance imaging mode exhibited higher accuracy in predicting SSC and anthocyanin content in blood oranges when compared to the diffuse reflectance mode. Among the tested conditions, preprocessing the original spectra with SNV and establishing a PLSR model utilizing full-wavelength spectrum yielded the highest prediction accuracy for SSC, where R pre was 0.927, RMSEP was 0.418 °Brix, and RPD was 2.621. On the other hand, preprocessing the original spectra with SNV and establishing a PLSR model with SPA-selected effective wavelengths exhibited optimal performance in predicting anthocyanin content, where R pre was 0.872, RMSEP was 1.702 mg/100 mL, and RPD was 1.918. Additionally, the spatial distributions of SSC and anthocyanin content in blood oranges were visualized using the optimal models. The findings demonstrate that hyperspectral imaging combined with effective spectral preprocessing and wavelength extraction algorithms can achieve non-destructive quality prediction of blood oranges.