Nondestructive measurement of sugar content of apple using hyperspectral imaging technique.

Abstract

Hyperspectral imaging technique is an upcoming and promising field of research for non- destructive quality assessment of agricultural and food products. It has a greater advantage of combining spatial imaging and spectral measurement which can detect both of the external and internal quality of the product. Sugar content is an important internal quality attribute for any fresh fruit. This research work focuses on evaluating the use of hyperspectral imaging technique which employs the wavelength range of 685-900 nm for detecting the quality of apple based on sugar content. The partial least square (PLS) method has the potential to produce the calibration and prediction model from their spectra. It was found that the optimal spectral range for sugar content of apple was 704.48-805.26 nm and the PLS calibration model for sugar content determination needed 4 PLS factors under standard normal variate (SNV) preprocessing method. The correlation coefficient (R) between the hyperspectral imaging prediction results and reference measurement results was equal to 0.90749. The PLS algorithm produced the calibration models which gave reasonably good correlation for estimating the sugar content of apple. It can thus be concluded that hyperspectral imaging technique is potentially useful for assessing sugar content of apple.