Determination of curcumin, starch and moisture content in turmeric by Fourier transform near infrared spectroscopy (FT-NIR)

Fourier transform near infrared spectroscopy (FT-NIR) in diffuse reflectance mode was used for the rapid estimation of curcumin, starch and moisture contents in turmeric samples. Thirty samples each of fingers and bulbs from varieties ‘Erode local’ and ‘Salem local’ (n = 120) were used for the study. Calibration models were developed and evaluated to describe the relationship between the three quality attributes with the NIR spectra of the turmeric powder. NIR reflectance spectra were acquired for each turmeric sample at a resolution of 8 cm⁻¹ over a wave number range of 12,500 to 3600 cm⁻¹. Vector normalization, first derivative and first derivative plus vector normalization were used as spectral pre-processing options. The relationship between the acquired spectra of turmeric samples and the quality attributes was examined through partial least square (PLS) regression algorithm. First derivative plus vector normalization technique predicted curcumin content with best accuracy with lowest root mean square error of cross validation (RMSECV) of 0.178% and maximum correlation coefficient for validation plots (R² = 91.9). Vector normalization technique predicted the starch and moisture content with RMSECV and R² value of 0.076%, 96.8 and 0.032%, 81.1 respectively. The results demonstrated that FT-NIR could be used as a rapid technique for quantification of curcumin, starch and moisture content in turmeric rhizomes for online grading in spice processing.