黑豆 (Glycine max L.) 富含蛋白質、碳水化合物和脂肪類,並含有礦物質、維生素、葉酸、花青素 (anthocyanin)、異黃酮、類黃酮、卵磷脂、酚酸等營養物質,為人們飲食中良好的營養成份來源。黑豆之初級壓榨黑豆油富含花青素、類胡蘿蔔素 (carotenoid) 和葉綠素 (chlorophyll) 等色素,這些色素對人體具有保健功效。目前黑豆油色素之相關研究極少,因此本研究擬透過利用高光譜吸收值與色素指數來建立簡單、快速並準確地測定黑豆油色素含量之預測模式。試驗材料之黑豆油以黑豆「台南3號」為原料,進行3種種皮添加和2種儲藏期間處理。測定黑豆油之高光譜吸收值以及花青素、類胡蘿蔔素、葉綠素a、葉黃素 (lutein)、玉米黃素 (zeaxanthin) 和葉綠素b含量。高光譜吸收值和色素指數與黑豆油中個別色素含量進行相關分析,並建立黑豆油色素預測模式。進一步將蒐集的9種黑豆油驗證樣品之高光譜吸收值和色素指數代入個別預測模式中,獲得各色素含量之預測值與實測值以殘差分析進行驗證。黑豆油花青素含量之最佳預測模式為高光譜吸收值A530和A649,差異色素指數DPI535, 524、DPI650, 575、DPI650, 588和DPI600, 710之預測模式。黑豆油類胡蘿蔔素含量之最佳預測模式為高光譜吸收值A440,差異色素指數DPI441, 524、DPI441, 575、DPI441, 588、DPI440, 710和DPI578, 710之預測模式。黑豆油葉綠素a含量之最佳預測模式為高光譜吸收值A450和A667,差異色素指數DPI668, 524、DPI430, 575、DPI667, 575和DPI667, 588之預測模式。葉黃素相對含量之最佳預測模式為高光譜吸收值A522和A547之預測模式。玉米黃素相對含量之最佳預測模式為高光譜吸收值A727之預測模式。葉綠素b相對含量之最佳預測模式為高光譜吸收值A641,差異色素指數DPI640, 575、DPI641, 588和DPI640, 710,比率色素指數RPI590, 575和RPI641, 710之預測模式
Black soybeans (Glycine max L.) are rich in protein, carbohydrates and fats, including minerals, vitamins, folic acid, anthocyanins, isoflavones, flavonoids, lecithin, phenolic acid, making them a good source of nutrients in daily diet. The first pressed black soybean oil is rich in pigments such as anthocyanins, carotenoids and chlorophyll, which have beneficial effects on human health. At present, there are only a few studies on black soybean oil pigment. Therefore, this research aims to establish a simple, rapid and accurate prediction model for the determination of black soybean oil pigment content by using hyperspectral absorbance and pigment index. The tested black soybean oil was obtained from black soybean ‘TN3’, with three seed coat additions and two storage periods. The hyperspectral absorbance of black soybean oil and the content of anthocyanin, carotenoid, chlorophyll a, lutein, zeaxanthin and chlorophyll b were measured. In order to establish the prediction model for the pigments, the hyperspectral absorbance and the pigment index were compared with the content of the individual pigments simultaneously and the correlations between them were found. The data collected from nine different black soybean oils were further applied into the prediction model. The verification of the prediction model was done by performing residual analysis, comparing the predicted and the measured content of each pigment. The optimal prediction model for anthocyanin is achieved by using the hyperspectral absorbance A530 and A649 and differential pigment index DPI535, 524, DPI650, 575, DPI650, 588 and DPI600, 710. As for the carotenoid, best prediction results were gained with the hyperspectral absorbance A440 and the differential pigment index DPI441, 524, DPI441, 575, DPI441, 588, DPI440, 710 and DPI578, 710. On the other hand, the best prediction of chlorophyll a were the ones with the hyperspectral absorbance A450 and A667 and the differential pigment index DPI668, 524, DPI430, 575, DPI667, 575 and DPI667, 588. As for the relative content of lutein, the optimal prediction outcome was obtained with hyperspectral absorbance A522 and A547. As for the relative content of zeaxanthin, the optimal prediction outcome was obtained with hyperspectral absorbance A727.The best prediction model for the relative content of chlorophyll b is with hyperspectral absorbance A641, differential pigment index DPI640, 575, DPI641, 588 and DPI640, 710, ratio pigment index RPI590, 575 and RPI641, 710.