近紅外光分析技術應用稻米品質之測定

Abstract

本研究利用由日本、美國及東南亞等地區收集之秈稻品種共24個，稉稻品種共29個為材料，在一期作於嘉義農業試驗分所試驗田種植，試驗主要目的在探討理化特性，諸如米飯物理特性（texture）、白米粉末黏度特性（viscosity）、貯藏性蛋白質及胺基酸等共40項理化特性與近紅外光吸收光譜之關係，並針對近紅外線分析技術（technique of near infrared reflectance spectroscopy，NIRS）推定米質之可行性進行探討，結果發現NIRS光譜吸收值與米理化特性間關係相當密切，利用19個NIRS波長吸光值轉化成19個主成份（principal component）後，再以19個主成份對各品質性狀進行逐步迴歸分析，結果於食味官能品評特性中，外觀、口味、黏度及食味總評之模式其複相關係數（R）分別為0.912、0.938、0.951 及0.961；而以6個物理特性建立之模式其複相關係數（R）介於0.558～0.819之間，其中黏性模式之解釋率最高（0.819）；12個黏度特性迴歸模式之複相關係數（R）介於0.554～0.940之間，其中以尖峰黏度之糊化始溫值（0.913）、破裂黏度（0.918）、相對破裂率（0.938）及破裂率（0.940）模式之解釋率較高，複相關係數（R）均達0.91以上；4種貯藏性蛋白質迴歸模式之複相關係數（R）介於0.63～0.869，其中以globulin迴歸模式解釋率較高（0.869）。18個胺基酸建立迴歸模式之複相關係數（R）介於0.565～0.894之間，其中除了色胺酸、異亮胺酸、甲硫胺酸及精胺酸等4種胺基酸之迴歸模式的複相關係數（R）低於0.80 外，其餘14種胺基酸之迴歸模式的複相關係數（R）均大於0.82。進一步以複相關係數（R）高於0.8之模式其米理化特性之預測值對食味總評值進行逐步迴歸分析，結果發現利用NIRS米理化特性預測值可解釋食味變異達91％。又利用NIRS 波長吸光值所轉化之主成份亦可解釋食味變異達92.4％，顯示以NIRS推定食味變異應是可行之途徑。 24 indica and 29 japonica rice varieties (53 in total) collected from Japan, America and southeast Asia were available in this study. By the courtesy of Chia-Yi Experiment Station, TARI all material were grown at that station in 1st crop season. The main objectives of this study were to determine the relationships between physicochemical properties and the absorbance of 19 NIRS wavelength. And to elucidate the feasibility for using near infrared reflectance spectroscopy (NIRS) on the assessment of rice grain quality. The results obtained by correlation analyses and stepwise regression analysis between the physicochemical properties and NIRS were summarized in the followings:The results were showed that the absorbance of 19 NIRS wavelength have close relationships with physicochemical properties, and the absorbance of 19 NIRS wavelength were transformed by principal components analysis and were regressed on the physicochemical properties by stepwise regression analysis. The results showed that the values (R) of multiple correlation coefficients were 0.912, 0.938, 0.951 and 0.961 for appearance, flavor, cohesion and overall sensory evaluation of palatability, respectively. The range of multiple correlation coefficients was 0.558~0.819 for texture traits of cooked rice. In this range, viscousness was found to be the highest (0.819) among the six texture traits. The values of multiple correlation coefficients ranged from 0.554 to 0.940 for 12 viscosity traits of milled rice and the gelatinization temperature of peak viscosity (0.913), breakdown (0.918), relative breakdown (0.938) and breakdown ratio (0.940) showed the highest four R values among the 12 viscosity traits. The values of multiple correlation coefficients ranged from 0.630 to 0.869 for 4 storage proteins of milled rice and the golbulin (0.869) showed the highest R value. Except tryptophan, isoleucine, methionine and arginine, all of the values of multiple correlation coefficients for other 14 amino acids were higher than 0.82. The predicted values of physicochemical properties in the models showing 0.8 or higher multiple correlation coefficients were selected to establish the palatability equation by stepwise regression analysis. The results indicated that the predicted values of physicochemical properties could explain 91% of variation of overall sensory evaluation. However, the principal components transformed from the absorbance of 19 NIRS wavelength could explain 92.4% of variation of overall sensory evaluation. These results showed that the NIRS might be used for determining the eating quality of cooked rice.