Non-destructive detection of soluble solids content in Shawo radish with spatial spectra extraction method based on the full transmission near-infrared spectroscopy

Abstract

Shawo radish valued for its unique flavor and nutritional properties, requires rapid and non-destructive soluble solids content (SSC) assessment for quality control. The inherent variability in Shawo radish length and heterogeneous internal compositional distribution hinder accurate prediction of both full-region and region-specific SSC using conventional spectroscopic techniques. This study developed a novel analytical approach based on near-infrared spectroscopy. Utilizing a self-developed short-integration full-transmission online device for dynamic multi-point spectral acquisition, a preprocessing method with spatial spectra extraction was introduced. This method minimized the interference from the regions with low SSC-correlation while enabling target spectral captured from specific regions. The effect of method was validated by comparing the models built with complete and extracted spectra. Optimization determined that extracting [5 %,15 %] of the total numbers of the spectra, combined with Savitzky-Golay Smoothing plus Standard Normal Variate (SGS+SNV), Competitive Adaptive Reweighted Sampling (CARS), and Partial Least Squares Regression (PLSR), yielded the optimal model for predicting the full-region SSC (R_p=0.894, $\mathrm{RMSEP}$=0.386°Birx, RDP=2.209). Furthermore, spatial spectra extraction successfully predicted SSC in the critical middle section of the radish (R_p=0.826, $\mathrm{RMSEP}$=0.520°Brix, RDP=1.680). These results demonstrate the spatial spectra extraction significantly enhances the accuracy of online dynamic SSC prediction for both full-region and region-specific quality assessment in Shawo radish.