This is the authors' abstract. We don't add key points for this paper.
Soluble solids content (SSC) is a critical indicator for evaluating Dingxiang pear quality and ripeness. Visible and near-infrared (400–1000 nm) hyperspectral imaging provides a nondestructive approach for SSC detection. However, the presence of uninformative variables and data redundancy in full-spectrum data restricts the efficiency of quantitative modeling. To address this limitation, this study investigated a two-stage variable selection strategy to optimize partial least squares calibration models. Specifically, interval-based methods (backward interval partial least squares, interval random frog, and interval variable iterative space shrinkage approach [iVISSA]) were coupled with wavelength-specific algorithms (competitive adaptive reweighted sampling and bootstrapping soft shrinkage [BOSS]) to identify the most informative variables. Results indicated that the iVISSA-BOSS model achieved the highest predictive accuracy, recording a root mean square error of prediction of 0.3013. The validation set’s coefficient of determination reached 0.8176, a marked improvement over the 0.7401 obtained by the full-spectrum model. These findings demonstrate that integrating continuous interval selection with discrete wavelength optimization effectively reduces data dimensionality and model complexity, providing an efficient algorithmic framework for the rapid quality assessment of Dingxiang pears.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Chu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: