Measurement of moisture and total reducing sugars using Near Infrared Spectroscopy

Abstract

Determining compound levels in fruits and vegetables is necessary in the food industry for appropriate management and processing of these products. A fast, nondestructive and reliable method of measuring compounds is extremely desirable for online industrial applications. In this project, the accuracy and feasibility of measuring moisture and total reducing sugar content in a vegetable medium using a Near Infrared Spectroscopy technique was investigated as an alternative to slow and tedious classical laboratory chemical analysis methods. Moisture reflectance spectra were analyzed in the 900-1600 nm region. Transmission spectra from sliced samples were tested within 2000 and 2500 nm. Using digital signal processing techniques such as multiplicative signal correction, and filtering coupled with statistical analysis tools such as partial least squares regression, the best prediction models were obtained with a standard deviation of 0.61 and 0.016 % for moisture and sugars, respectively. The correlation coefficient was 0.95 for moisture. The corresponding value for sugar was 0.975. Best results were obtained when models were created using single vegetable cultivars in calibrating the prediction models. This study showed that when combined cultivars are used in calibrating the prediction models, the predictive capability is below the required standards. Overall, the results showed that measurements using Near Infrared Spectroscopy are feasible in the applications investigated; however, these results are in the preliminary stages and more research needs to be conducted to improve the measurement accuracy, and further develop the technique before indusuw implementation.