Non-destructive assessment of apple internal quality using rotational hyperspectral imaging

• DOI: 10.3389/fpls.2024.1432120
• Publication Date: 2024-11-06T12:00:49Z
• Authors: Xiaojiang Wang; Junying Han; Chengzhong Liu; Tong Feng
• Keywords: Titratable acid; Principal component regression

This work aims to predict the starch, vitamin C, soluble solids, and titratable acid contents of apple fruits using hyperspectral imaging combined with machine learning approaches. First, a camera by rotating samples was used obtain images fruit surface in spectral range 380~1018 nm, its region interest (ROI) extracted; then, optimal preprocessing method preferred through experimental comparisons; on this basis, genetic algorithms (GA), successive projection (SPA), competitive adaptive reweighting adoption (CARS) were extract feature variables; subsequently, multiple models (support vector regression SVR, principal component PCR, partial least squares PLSR, linear MLR) model inversion between internal nutrient quality physicochemical indexes fruits, respectively. Through comparative analysis all prediction results, it found that among them, for content, 2 nd Der-CARS-MLR superior performance (the coefficients determination R p exceeded 90% them). In addition, potential relationships four nutritional qualities explored based t-values p-values, significant conclusion drew starch C highly correlated.