透過您的圖書館登入
IP:18.218.129.100
  • 期刊
  • OpenAccess

FEASIBILITY OF NEAR-INFRARED INSPECTION TO ASSESS THE PROTEIN CONTENT OF SINGLE RICE KERNELS

近紅外光偵測單粒白米蛋白質可行性評估

摘要


The main objective of this research was to investigate the feasibility of using a low-cost near-infrared inspection system to rapidly determine the protein content of single rice kernels. We used a near-infrared system with low-cost band-pass filters and varying wavelengths of 890, 910, 920, 990, and 1,014 nm to inspect single rice kernels. We then applied a fuzzy inference system (FIS) to convert measurement values into protein content values for individual kernels. The experimental results were verified by comparing the predicted protein quantities with standard values obtained using a spectrophotometer at a wavelength below 595 nm. The average and median error rates of the proposed rapid inspection technique were 10% and 5%, respectively. The near - infrared inspection system has a processing capacity of 730 samples per second per chute channel. We concluded that this system could be utilized with a high processing capacity to perform quality inspection of single rice kernels.

並列摘要


本研究主要目的在於發展一套低成本近紅外光檢測系統,用於快速檢測單粒白米蛋白質含量。近紅外光檢測系統結合光譜影像及近紅外光分光帶通濾鏡分別使用890、910、920、990及1,014 nm不同波長針對單粒白米進行檢測試驗。應用模糊推論系統(Fuzzy inference system)將近紅外光檢測系統量測值轉換為單粒白米蛋白質的含量值,並利用分光光譜儀於595 nm波長下所獲得之標準值,與其進行單粒白米蛋白質含量預測驗證。實驗結果顯示,近紅外光檢測系統的平均誤差和中位誤差率分別為10%和5%,且近紅外光檢測系統的槽道處理能力每秒可為730顆樣本,系統具有高效率單粒白米蛋白質含量檢測之處理能力。

延伸閱讀