Non-destructive measurement of sugar content in Fuji apple with bifurcated fiber optic sensor

To develop a nondestructive sugar analyzer for intact apples, the potential of Fourier Transform Infrared (FTNIR) method with bifurcated fiber optic sensor was evaluated. Three different kinds of mathematical treatments (original, first derivative and second derivative) in range of 800-2500nm were discussed. A total of 120 Shandong Fuji apples were tested and 80 of them were used to form a calibration data set. The relationship was established between the diffuse reflectance spectra and the sugar content by means of the partial least squares analysis (PLS) technique. The influence of the data preprocessing was investigated and the optimal wavelength range was also found in the range of 967-1831nm. Depending on data preprocessing and PLS analysis, three predictive models had a correlation coefficients of 0.97, 0.95 and 0.97 with a ratio of data standard deviation to the root mean square error of prediction (SDR) of 3.18 (>3.00), 2.55(<3.00) and 3.23 (>3.00) for original, first derivative and second derivative of spectra respectively; 3.00 was considered the minimum ratio value for only sorting fruit. The results show that the second derivative spectra data gave the best prediction result. It is concluded that the FTNIR method with bifurcated fiber optic sensor yields an accurate estimate of the sugar content in intact apples.