초록
도정하지 않은 정조의 81 시료로부터 스펙트럼을 수집하고, 백미 완전미도정수율 예측 희귀모델을 개발하기 위해 검량식을 작성한 결과 스펙트럼을 8 nm 간격으로 지정하고, 1차미분 방법으로 검량식을 작성한 완전미율의 결정계수는 MPLS에서 0.8353, PLS 방법에서 0.8416, PCR에서 0.5277를 나타냈다. 스펙트럼을 20 nm 간격으로 지정하고 1차미분 방법으로 검량식을 작성하였다. 완전미율의 결정계수는 MPLS에서 0.8144, PLS 방법에서 0.8354, PCR에서 0.6809를 나타냈다. 스펙트럼을 8 nm 간격으로 지정하고 2차미분 방법으로 검량식을 작성하였다.완 전미율의 결정계수는 MPLS 방법에서 0.7994, PLS에서 0.8017, PCR에서 0.4473을 나타냈다. 스펙트럼을 20 nm 간격으로 지정하고 2차미분 방법으로 검량식을 작성하였다. 완전미율의 결정계수는MPLS 방법에서 0.8004, PLS에서 0.8493, PCR에서 0.6609을 나타냈다.
The purpose of this study was to measure fundamental data required for the prediction of milling ratios, and to develop regression models to predict the head rice ratio of milled rice using NIR spectra of hulled rice. A total of 81 rice samples used in this study were collected from Jeongeup, Jeonbuk province in 2006. NIR spectra were measured using one mode of measurement, reflection. The reflectance spectra were measured in the wavelength region of 400-2500 nm with an NIR spectrophotometer "NIRSystems 6500" (Foss, Silverspring, USA). Calibration equations were developed by the modified partial least squares (MPLS), partial least squares (PLS), and principal components regression (PCR). Math treatments were 1-4-4-1, 1-10-10-1, 2-4-4-1, and 2-10-10-1. The software used was WinISI (Infrasoft International, State College, USA). Automatic head rice production and quality checking system used was "SY2000-AHRPQCS" (Ssangyong, Korea). The calibration was made with the first derivative and the spectrum designated was in 8 nm interval. The determination coefficients of head rice ratios were 0.8353, 0.8416 and 0.5277 for the MPLS, PLS and PCR, respectively. Those obtained with 20 nm interval were 0.8144, 0.8354 and 0.6908 for the MPLS, PLS and PCR, respectively. The calibration was made with second derivative that spectrum designated was 8 nm in interval. The determination coefficients of head rice ratios were 0.7994, 0.8017 and 0.4473 for the MPLS, PLS and PCR, respectively. Those with 20 nm interval were 0.8004, 0.8493 and 0.6609 for the MPLS, PLS and PCR, respectively. These results indicate that the accuracy of determination coefficient for MPLS and PLS is higher than that of PCR.
