Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/35767
標題: 不同照蛋光源對種蛋孵化影像之相關性研究
The Effects of Different Candling Light Sources on Hatching Images
作者: 曾國豪
Zeng, Guo-Hao
關鍵字: 種蛋;Hatching;孵化;機器視覺影像;Incubation;Machine vision
出版社: 生物產業機電工程學系所
引用: 1. 李殿奎編著。1998。在光學世界裡。凡異出版社。 2. 夏良宙。1995。影響種雞蛋孵化率的營養因子。飼料營養雜誌。九五年第十期:p.23~29。 3. 馬春祥編著,1980。家禽學上冊,國立編譯館主編,黎明文化事業股份有限公司出版。179-192。 4. 張善勝編著。1999。蛋品加工學。華香園出版社。 5. 陳加增。2000。近紅外光應用於水果糖酸度線上檢測之研究。碩士論文。台北:國立台灣大學農業機械工程學系碩士班。 6. 童連編著。2010。家禽孵化技術。五洲出版社。27-35;181-202。 7. 楊清富。1993。應用機器視覺進行蕃茄顏色分級之研究。碩士論文。台中:國立中興大學農業機械工程研究所。 8. 董子毅,2008。危險評估中ROC 曲線在預測2×2表上與敏感度及特異度之關係。亞洲家庭暴力與性侵害期刊。第四卷第二期,2008,64 -74 頁。 9. 劉志春編著。2013。雞類飼養與繁殖。五洲出版社。42-46。 10. 鄭人嘉。2004。影像處理技術應用於種蛋孵化過程監測之研究。碩士論文。台中:國立中興大學生物產業機電工程學研究所。 11. 繆紹綱編著。2004。數位影像處理。普林斯頓國際有限公司。 12. 農業統計年報。2012。行政院農業委員會。 13. 孵化技術手冊。1986。台灣省家禽孵化商業同業公會聯合會。
摘要: 
The purpose in this study is to investigate the effect of hatching images for Machine vision select with different color of hatching lights. The subjects are the eggs of Lohmann fowl. The first of all is to investigate the difference between the lateral and bottom side of hatching images which are candled by the side of air cell. Hatching lamps are associated with web cameras to capture the lateral and bottom side of hatching images daily from day 1 to day 20, and subsequently perform image processing to obtain the RGB and HSI color models of normal, unfertilized and suspended eggs. The result shows that taking image on the bottom side and the lateral side provide the same determination.
In this study, five different colors of LED light bulbs are chosen as candling light sources, and control the environment of the incubator under the temperature of 100℉ and the humidity of 80%. Hatching lamps are associated with web cameras to capture the lateral side of hatching images daily, and convert the hatching images into RGB image gray values. A static method, namely Receiver Operating Characteristic (ROC) curve, is applied to find best judgment tangent point by daily experimental group that will be verified with validation group.
In the result, the G value of yellow light to best determine the unfertilized and early suspended eggs. On the sixth day, the accuracy for determining the unfertilized eggs is 80.6%, the accuracy of determining the fertilized eggs is 100%. On the ninth day, the accuracy for determining the early suspended eggs (including unfertilized eggs) is 98.6%, the accuracy of determining the poultry eggs (including later suspended eggs) is 100%.

本研究主要的目的在探討在以不同照蛋燈之顏色進行照蛋,其種蛋孵化影像對於機器視覺選別之影響。研究對象為龍門蛋雞之種蛋,首先探討以氣室端照蛋後種蛋側面及底面影像擷取之差異,每日以照蛋燈配合網路攝影機,擷取側面與底部第1至20天之孵化影像,並進行影像處理取得出雛蛋、無精蛋與初期中止蛋之RGB與HSI色彩模型其結果可以發現底部取像結果與側面取像結果有相同的判斷力,RGB模型在判斷比HSI模型有較佳的判斷力。
本研究利用5種不同燈色之LED燈泡作為照蛋光源並將孵化環境控制在溫度100℉、相對濕度80%,每日以照蛋燈配合網路攝影機,擷取底面種蛋影像並轉換為RGB灰階值後利用統計方法ROC曲線,以實驗組建立每日之最佳的判斷切點,並以驗證組進行驗證。結果顯示以黃光照蛋利用的G值進行判斷為最佳,在第六天無精蛋準確率為80.6%,受精蛋準確率為100%。於第九天初期中止蛋(含無精蛋)準確率為98.6%,出雛蛋(含後期中止蛋)準確率為100%。而過了第二死亡高峰期的後期中止蛋,無法藉由影像判斷出差異。
URI: http://hdl.handle.net/11455/35767
其他識別: U0005-2608201322571500
Appears in Collections:生物產業機電工程學系

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.