Flaxseed protein content prediction based on hyperspectral wavelength selection with fractional order ant colony optimization

• DOI: 10.3389/fnut.2025.1551029
• Publication Date: 2025-04-15T04:12:20Z
• Authors: Bo Wang; Junying Han; Chengzhong Liu; Jianping Zhang; Yanni Qi
• Keywords: VNIR; Content (measure theory)

The protein content of flaxseed ( Linum usitatissimum ) is a crucial factor influencing its nutritional value and quality. Spectral technology combined with advanced modeling methods offers fast, accurate, cost-effective approach for predicting content. In this study, visible-near infrared hyperspectral imaging (VNIR-HIS) was fractional order ant colony optimization (FOACO) to determine the flaxseed. Thirty varieties commonly cultivated in Northwest China were selected, data along measurements collected. A joint x-y distance algorithm applied divide dataset into calibration prediction sets after removing outliers. Partial least squares regression (PLSR) models developed based on both raw preprocessed spectra, Savitzky-Golay (SG) smoothing method found provide superior performance. performance wavelength selection FOACO, principal component analysis (PCA), (ACO) compared using PLSR multiple linear (MLR) models. FOACO-MLR model achieved accuracy 0.9248, root mean square error (RMSE) 0.4346, relative deviation (RPD) 3.6458, absolute (MAE) 0.3259. results show that provides significant advantages content, particularly terms stability characteristic bands. By combining VNIR-HIS FOACO algorithm, study an efficient rapid determining flaxseed, providing reliable technical support precise detection components.