Abstract:Starch food is easy to retrograde during processing, transportation and storage, and the degree of retrogradation seriously affects the nutritional value and shelf-life of starch food. Soretrogradation degree is really expected to determine rapidly and non-destructively during storage, that is near-infrared and mid-infrared spectroscopy. The near-infrared and mid-infrared spectra of starch in different storage times (0d, 1d, 2d, 3d, 4d, 5d, 10d, 15d and 20d) were collected. There was a certain associations between spectra data and chemical reference detected by spectrophotometry, then chemometrics (partial least squares, PLS) were used to establish the prediction model of starch retrogradation with near-infrared, mid-infrared and fusion data, the best one that had higher correlation coefficient and lower error was chosen. The results showed that the backward interval partial least squares (biPLS) prediction model of fusion technology was the best one, the root mean square error of cross-validation (RMSECV) and root mean square error of prediction (RMSEP) were 6.79% and 9.52%, and the calibration and prediction correlation coefficient were 0.9655 and 0.9313, respectively. The results indicated that the fusion spectroscopy was superior to any single spectral technique, which could provide more accurately information of starch. Hence, the infrared spectroscopy could detect the retrogradation degree of corn starch rapidly and non-destructively, provide guidance for the processing of starchy food, and ensure the quality and safety of starchy food.