臺灣博碩士論文加值系統

本研究建構一套水稻種子辨識輔助系統，其功能包括自動進料、計數、擷取影像、特徵萃取、品種辨識、建置資料庫與出料，並應用此系統針對臺灣五大水稻種子(台梗9號、台南11號、台梗14號、台中192號與台中秈10號)進行品種辨識。
本研究利用影像處理技術針對水稻種子外觀與顏色進行形態特徵與色彩特徵之萃取。首先應用所萃取之形態特徵建置形態貝氏分類器並針對台中秈10號與其餘4種品種(台梗9號、台南11號、台梗14號與台中192號)進行分類。接者，利用序列浮動向前選擇法針對所萃取之特徵進行挑選，再應用所挑選出之特徵進行色彩貝氏分類器之建置並針對台中192號與其餘3種品種(台梗9號、台南11號與台梗14號)進行分類；最後，利用形態特徵建置類神經網路分類器並針對台梗9號、台南11號與台梗14號進行分類。
本研究建構之水稻種子辨識輔助系統之選別速率為74顆/分，針對不同年度之原原種與原種之水稻種進行測試，測試結果為2014年原原種之平均選別率與平均影像辨識率分別為90.71%與95.87%；2014年原種之平均選別率與平均影像辨識率分別為77.86%與80.59%；2015年原原種之平均選別率與平均影像辨識率分別為80.83%與85.27%。

In this study, an image identification assistance system for paddy seed was developed. The functions included auto-feeding, counting, image capture, variety identification and establishment of image database, Taiwan five varieties of paddy seed were identified by this system.
In this study, image processing techniques were used to extract shape features and color features based on appearance of paddy seeds. Firstly, shape Bayesian classifier was established by two of shape features to classify Taichung sen 10 and other 4 varieties including Taikong 9, Tainan 11, Taikong 14 and Taichung 192. Secondly, color Bayesian classifier was established by optimal feature combination which was selected by sequential floating forward selection to classify Taichung 192 and other 3 varieties including Taikong 9, Tainan 11 and Taikong 14. Finally BPNN classifier was established by 10 shape features to classify Taikong 9, Tainan 11 and Taikong 14.
The classification rate of image identification assistance system for paddy seed was 74seeds / min. Stock and foundation paddy seeds in different year were presented to test the system. The average classification accuracy and image identification of foundation paddy seed was 90.71% and 95.87% in year 2014; the average classification accuracy and image identification of stock paddy seed was 77.86% and 80.59% in year 2014; the average classification accuracy and image identification of foundation paddy seed was 80.83% and 85.27% in year 2015.

摘要 i
Abstract ii
目錄 iii
圖目錄 v
表目錄 viii
第一章 前言 1
1.1 研究動機與目的 1
1.2 論文架構 2
第二章 文獻探討 3
2.1 水稻種子外觀概述 3
2.2 水稻種子品種介紹 4
2.3 水稻種子檢查介紹 6
2.4 文獻探討 8
第三章 實驗理論分析 12
3.1 影像處理基本理論 12
3.2 Otsu法之自適應門檻值 16
3.3 邊界鏈碼追蹤法 16
3.4 凸包演算法 18
3.5 橢圓傅立葉描述子 21
3.6 多項式曲線擬合 25
3.7 序列浮動向前選擇 26
3.8 倒傳遞類神經網路 28
3.9 貝氏分類器 30
第四章 水稻種子特徵之估算 32
4.1 影像分割 32
4.2 定位點之找尋 35
4.3 帶梗樣本之辨識 37
4.4 特徵萃取 38
4.5 龍骨特徵之萃取 43
第五章 水稻種子辨識輔助系統之架構 49
5.1 實驗樣本 49
5.2 實驗設備 51
5.3 水稻種子辨識輔助系統之軟體 54
5.4 水稻種子辨識輔助系統之動作流程 57
第六章 結果與討論 59
6.1 水稻種子辨識輔助系統測試 59
6.2 特徵選取結果 64
6.3 分類器建置結果 74
6.4 水稻種子辨識結果 77
第七章 結論與未來展望 90
7.1 結論 90
7.2 未來展望 91
參考文獻 92
附錄A 水稻種子辨識輔助系統之硬體規格 95
附錄B PLC I/O 97

[1]黃卯昌、許鐈云、張仁銓、周永吉、黃亮白、蘇士閔、陳易徵、洪建民、黃玉梅。種種守護-臺灣水稻種子檢查介紹。種苗科技專訊。2013。84:18-21。
[2]種苗改良繁殖場種子檢查室。102年米檢人員講習。2013。
[3]International Seed Testing Association (ISTA). International Rules for Seed Testing. Vol. 2015.
[4]侯福分。優良水稻栽培管理技術手冊。2009。行政院農業委員會農糧署。
[5]戴振洋、張致盛。臺中區農業改良場歷年育成品種專輯。2009。行政院農業委員會台中區農業改良場。
[6]陳素娥、林孟輝。有機水稻專輯。桃園區農業專訊。2003。62:12-18。
[7]Chaugule A., S.N. Mali. Evaluation of Texture and Shape Features for Classification of Four Paddy Varieties. Hindawi Publishing Corporation Journal of Engineering 2014, 1-2
[8]Gujjar Harish S., Dr. M. Siddappa. A Method for Identification of Basmati Rice grain of India and Its Quality Using Pattern Classification. International Journal of Engineering Research and Applications 2013, 3(1): 268-373
[9]MousaviRad S.J., K. Rezaee, K. Nasri. A new method for identification of Iranian rice kernel varieties using optimal morphological features and an ensemble classifier by image processing. Majlesi Journal of Multimedia Processing 2012, 1(3): 1-8.
[10]Mebatsion H.K., J. Paliwal, D.S. Jayas. Automatic classification of non-touching cereal grains in digital images using limited morphological and color features. Computers and Electronics in Agriculture 2012, 90(2013): 99–105.
[11]Wiwart M., E. Suchowilska, W. Lajszner, Ł. Graban. Identification of hybrids of spelt and wheat and their parental forms using shape and color descriptors. Computers and Electronics in Agriculture 2012, 83(2012): 68–76.
[12]Szczypinski P.M., A. Klepaczko, P. Zapotoczny. Identifying barley varieties by computer vision. Computers and Electronics in Agriculture 2015, 110 (2015): 1–8.
[13]廖家興。機器視覺在糙米外觀檢測之應用。碩士論文。1995。台中：國立國立中興大學農業機械工程研究所。
[14]林建銘。稻米品質檢測軟體之設計與檢測系統之改進之研究。碩士論文。1999。台中：國立國立中興大學農業機械工程研究所。
[15]Sun T.H., F.C. Tien, R.J. Kuo. Automated thermal fuse inspection using machine vision and artificial neural networks. Journal of Intelligent Manufacturing 2014, 902-915.
[16]Lee C.Y., L. Yan, T.F. Wang, S.R. Lee, C.W. Park. Intelligent classification methods of grain kernels using computer vision analysis. Measurement Science and Technology 2011, 22(6): 64006-64012.
[17]Liu Z.Y., F. Cheng, Y.B. Ying, X.Q. Rao. Identification of rice seed varieties using neural network. Journal of Zhejiang University SCIENCE 2005, 6B(11):1095-1100.
[18]黃國益。蝴蝶蘭大苗選別系統之開發。博士論文。2002。台中：國立國立中興大學農業機械工程系。
[19]蔡瑞平。濕稻穀未成熟粒檢測系統之研製。碩士論文。2013。台北:華梵大學機電工程系。
[20]Laursen M.S., H.S. Midtiby, N. Krüger, R.N. J?rgensen. Statistics-based segmentation using a continuous-scale naive Bayes Approach, Computers and Electronics in Agriculture 2014, 109(2014): 271–277.
[21]Gonzalez R.C., R.E. Woods, Digital Image Processing 3rd ed. Pearson Education Ltd.
[22]Cormen T.H., C.E. Leiserson, R.L. Rivest, C. Stein, Finding the convex hull. Introduction to Algorithms, Second Edition. 2001. 33(3): 826-833.
[23]Kuhl F.P., C.R. Giardina. Elliptic Fourier Feature of a Closed Contour. Computer Graphic and Image Processing 1982, 18: 236-258.
[24]Myers R. H. Classical and Modern Regression with Application 2nd Edition. Cengage Learning, Inc.
[25]Pudil P., J. Novovicova, J. Kittler, Floating search methods in feature selection. Patter Recognition Letters 1994, 15: 1119-1125.
[26]Hangan M.T., H.B. Demuth, O.D. Jesus. Neural Network Design 2nd Edition. Taipei: Thomson Learning.
[27]Fukunaga K., Introduction to statistical pattern recognition Second Edition. Academic Press, Inc.
[28]Duda R.O., P.E. Hart, D.G. Stork. Pattern Classification. John Wiley & Sons, 2nd edition.
[29]Choudhury S.D., T. Tjahjadi. Silhouette-based gait recognition using Procrustes shape analysis and elliptic Fourier descriptors. Pattern Recognition 2011, 45(2012): 3414–3426.