Assessing Moisture Content of Coolstored Blueberries During Water Loss Scenarios Using Hyperspectral Imaging

Water loss is a key factor affecting the postharvest quality and shelf life of blueberries, and storage conditions (humidity and time) play an important role in regulating water retention capacity of stored berries. This study aims to explore the variation of moisture content (MC) in blueberries under different storage humidity and storage time conditions and to construct a prediction model for MC based on hyperspectral imaging. Four storage times (0, 7, 14, and 21 days) and three relative humidity (RH) levels (96%, 75%, and 56%) were set up in the experiment, and a hyperspectral imaging (HSI) system was used to collect image data of blueberries stored under different treatment conditions at 1.5°C. After preprocessing the spectral data using techniques such as denoising, normalization, and absorbance conversion, the partial least squares regression (PLSR) method was used to establish a water content prediction model. The results show that the model developed using data from a specific storage time has higher prediction accuracy than that from across the various storage periods. X‐loadings analysis shows that the 652–699 nm and 961–984 nm bands in the Vis‐NIR range have significant contributions to the MC estimation model. The best‐performing model was obtained for berries stored for 21 days at varying RH conditions (R 2 = 0.74; root mean square error (RMSE) = 1.0% MC). Additionally, using the same spectral input data, compression firmness of the berries after 21 days of coolstorage could be estimated using PLS models with reasonable accuracy (R 2 = 0.67; RMSE = 0.47 N).