Estimating the starch-iodine Blue Value of Residual Liquid of Rice by near-infrared Reflectance Spectroscopy

利用近紅外線反射光譜估計稻米炊飯液碘呈色度

10.30089/JAF.200109.0006

郭寶錚(Bo-Jein Kuo)；洪梅珠(Mei-Chu Hong)

炊飯液碘呈色度 ； 近紅外線反射光譜 ； 共線性 ； 複線性迴歸 ； 主成份迴歸 ； 淨最小平方迴歸 ； starch-iodine blue value of residual liquid BV ； NIRS ； multicollinearity ； MLR ； PCR ； PLSR

農林學報

50卷3期（2001 / 09 / 01）

67 - 77

英文

炊飯液碘呈色度此一特性可被用來預測稻米的嗜口性。本研究的目的在於評估利用近紅外線反射光譜研究估計稻米炊飯液碘呈色度的可能性並比較複線性迴歸，主成份迴歸及淨最小平方法在建立模式及預測能力上的表現。228個水稻樣品包括秈、稉及糯稻被用來分析炊飯液碘呈色度，並建立檢量線，結果發現利用4個波長的複線性迴歸模式可產生最高的相關係數及最低的預測機差，也就是說其在建立模式及預測能力上表現最佳。利用17個主成份所構成的主成份迴歸模式又較6個成份所構成的淨最小平方迴歸模式為佳，但是複線性迴歸中IRV值(82.042)遠大於主成份迴歸(18.331)及淨最小平方迴歸(8.356)的IRV值。顯示也線性迴歸模式較其他模式包含更多的隨機干擾。綜合言之，就本研究得知，稻米炊飯液碘呈色度的檢量線並不夠準確，在水稻育種上近紅外線反射光譜對稻米炊飯液中碘呈色度僅能做初步估計之用。

Starch-iodine blue value of residual liquid (BV) could be used to predict the palatability of rice. The objectives of this study were to evaluate the potential of Near-infrared ref1ectance spectroscopy (NIRS) to estimate BV of rice and compare the performance of model-building and prediction in developing the calibration equation using multiple linear regression (MLR), principal component regression (PCR), and partial least squares regression (PLSR), respectively. A total of 238 rice samples, including indica, japonica, and waxy rices, were employed in analyzing the BV. The 4-wavelength MLR model gave the highest correlation coefficient and the lowest standard error of prediction than PCR and PLSR models. In other words, This MLR model had the best predictive ability in estimating the BV of rice. The predictive ability of the PCR model with 17 principal components was slightly better than that of the PLSR model with 6 components. However, The IRV value in MLR model (82.042) was much larger than that in PCR (18.331) model and PLSR (8.356) model, suggesting that more random noise was involved in MLR model than that in other models. General1y speaking, the performances of model-building and prediction demonstrate that calibration for BV is less accurate and NIRS can be used only for the first approximation of BV in the rice breeding program.