Assessment of Egg Freshness by Using Transmission Spectral and Haugh Unit

根據行政院農委會統計，零售蛋佔國內雞蛋市場相當大的比例，隨著近年來食安意識的上漲，零售蛋的品質管理也應受到重視，故本研究以光譜響應數據和豪氏單位、蛋黃係數、蛋黃及蛋白pH值進行多元線性迴歸分析，建立相關之檢量線，以進行蛋品品質評估。 由養雞場購入當日產出未受精之白殼蛋及紅殼蛋，並置於常溫常濕之儲存環境，拍攝光譜響應數據後立即破蛋，以獲得欲量測之蛋品品質參數，共計白殼蛋及紅殼蛋各300顆，對光譜響應數據進行標準化及標準常態變數（Standard Normal Variate, SNV）等光譜預處理，以多元線性迴歸（Multiple Linear Regression, MLR）建立相關之檢量線並比較其預測能力，另取市售之白殼及紅殼洗選蛋各60顆評估檢量線對其之預測能力。結果顯示，白殼蛋及紅殼蛋之四項蛋品品質參數皆以SNV處理後所建立之檢量線有較佳的預測能力，其中以對豪氏單位之檢量線有最高的判定係數（coefficient of determination, r2），白殼蛋之校正組和預測組r2分別為0.812和0.804；紅殼蛋之校正組和預測組r2分別為0.806和0.807。According to the statistics of Council of Agriculture, Executive Yuan, retail eggs account for a large proportion of the domestic egg market. With the rising awareness of food safety in recent years, the quality management of retail eggs should also be valued. Therefore, this study attempted to use transmission response data and Haugh unit, Yolk index, Yolk and Albumen pH with multiple linear regression, and established relevant calibration curves for egg quality assessment. The unfertilized white shell eggs and brown shell eggs were purchased on the day from the chicken farm, each color was 300 and placed in room temperature and humidity storage environment. After the transmission response data was taken, immediately the egg's shell was peeled to obtain the egg quality parameters to be measured. Transmission response data were subjected to spectral preprocessing, such as traditional normalization and Standard Normal Variate (SNV). Multiple linear regression (MLR) was used to establish the relevant calibration curves and compared the predicted capabilities. In addition, white shell and brown shell washed eggs were 60 respectively, used to test calibration curves. The result showed that the calibration curves established after SNV processing has better predicted capabilities. Calibration curves for Haugh unit had the highest coefficient of determination (r2). The r2 for calibrated group and predicted group of white shell eggs were 0.812 and 0.804；For brown shell eggs were 0.806 and 0.807.目錄 謝誌 i 摘要 ii Abstract iii 目錄 iv 圖目錄 vii 表目錄 xi 第一章 前言 1 1.1研究背景 1 1.2研究目的 3 第二章 文獻探討 4 2.1非破壞光譜檢測技術之應用 4 2.2雞蛋品質檢測技術 6 2.3光譜預處理及檢量線建立 11 第三章 理論分析 13 3.1光學頻譜分析 13 3.2光譜預處理技術 16 3.2.1比爾-伯朗定律 16 3.2.2微分光譜 16 3.2.3比例縮放 17 3.3雞蛋隨儲存時間增加產生之物質 18 3.4蛋品品質 19 3.4.1 豪氏單位 19 3.4.2 蛋黃係數 20 3.4.3 蛋白及蛋黃pH值 21 3.5迴歸分析 22 第四章 實驗材料與方法 24 4.1 實驗材料 24 4.2 實驗設備 25 4.2.1光譜檢測設備 25 4.2.2蛋品品質量測與影像擷取 27 4.3實驗方法 30 4.3.1實驗設計 30 4.3.2光譜預處理及迴歸分析 33 4.3.3蛋品品質參數量測 34 4.3.4實驗流程圖 35 第五章 結果與討論 36 5.1 光譜響應變化 36 5.1.1原始光譜響應數據變化 36 5.1.2單一樣品重複檢測結果 38 5.2 蛋品品質逐日變化 42 5.3 多元線性迴歸分析 49 5.3.1光譜響應與豪氏單位之關係 49 5.3.2光譜響應與蛋白pH值之關係 61 5.3.3光譜響應與蛋黃係數之關係 72 5.3.4光譜響應與蛋黃pH值之關係 81 5.4 檢量線對市售洗選蛋之預測能力評估 82 第六章 結論與建議 84 6.1 結論 84 6.2 建議 86 參考文獻 8