基於形態學和貝氏分類器，本論文提出西瓜農地空拍影像辨識及盤點西瓜產量的有效方法。在大約6公頃西瓜農地的空拍影像中，充斥帆布、藤蔓、廢棄小西瓜、石塊等許多雜亂的信息，會嚴重干擾西瓜的辨識及盤點結果。
本研究透過訓練貝氏分類器模型，得到系統預備知識後，分為三個步驟：首先，從RGB圖像的G通道，計算大津閥值獲得二值化影像。爾後，用數理形態運算去除影像中的藤蔓。為確保西瓜的品質，農民實施一株一果並在農地遺留廢棄的小西瓜會影響產量盤點的精確性。數理形態運算後影像中保留的物件物為西瓜和帆布，透過計算每個物件物的面積，將小面積的物件物除去，去除的小物件物可能是廢棄小西瓜或小區塊的帆布。最後，以貝氏分類器檢測器分類帆布和西瓜。
根據實驗結果，貝氏分類器的精確度(precision)為95.51％，優於未使用貝氏分類器；6公頃的空拍影像約需10分鐘可盤點西瓜的產量。產量估算的準確度(F1 Score)為90.08％，和專家討論後此結果為可接受的。

Based on morphology and Bayesian classifier, this study proposes an efficient method to recognize and estimate the yield of watermelon from air borne image of watermelon field. In the air borne image, approximately six-hectares fields are covered in the image, much noisy information such as canvas, vine, small watermelon, etc. could interfere the outcome of recognition.
Therefore, this study based on training Bayesian classify model to obtain system prior knowledge, then employs three steps to overcome the problem. First, the binary image through calculating Otsu threshold from G channel of the RGB image. Second, the mathematical morphology is used to remove the information of vine. Due to keep the quality of watermelons, farmers generally adopt one plant to bring one fruit. The watermelon and canvas will reserve in the image and be the target. Calculating each area of target and removing the small target, that might be small watermelon or small canvas. Third, Bayesian classifier is used to distinguish canvas and watermelon.
The experimental results show that, the recognized precision with Bayesian classifier is 95.51%, it is better than without Bayesian classifier. The execution time requires 10 minutes for six-hectares fields air borne image to estimate the yield of watermelon. The accuracy (F1 Score) of yield estimation is 90.08% and acceptable by experts.

第一章　緒論　1
1-1　前言　1
1-2　文獻回顧　2
1-3　研究動機與目的　3
1-4　研究方法與貢獻　5
1-5　論文架構　10
第二章　相關背景知識　12
2-1　影像前處理　12
2-2　數理形態學　17
2-3　貝氏定理及貝氏分類器　19
第三章　西瓜辨識及數量盤點系統　24
3-1　系統流程圖　24
3-2　貝氏分類器模型　26
3-3　系統演算法　33
第四張　實驗結果　41
4-1　實驗結果　41
4-2　結果分析　54
第五章　結論與未來工作　66
5-1　結論　66
5-2　未來工作與未來展望　67

[1] 行政院農委會https://www.coa.gov.tw/ws.php?id=2505139, 6, Aug. 2019.
[2] T. T. Nguyen, V. N. Hoang, T. L. Le, Thanh-Hai Tran, and H. Vu “A Vision Based Method for Automatic Evaluation of Germination Rate of Rice Seeds,” 2018 1st International Conference on Multimedia Analysis and Pattern Recognition (MAPR),pp.1-6, Ho Chi Minh City, Vietnam, Apr. 2018.
[3] K.S. Jamuna, S. Karpagavalli, M.S. Vijaya, P. Revathi, S. Gokilavani, and E. Madhiya, “Classification of Seed Cotton Yield based on the Growth stages of Cotton crop using Machine Learning Techniques,” 2010 International Conference on Advances in Computer Engineering, pp.312-315, Bangalore, India, June 2010.
[4] A. Zhu and L. Yang, “An improved FCM algorithm for ripe fruit image segmentation,” 2013 IEEE International Conference on Information and Automation, pp. 436-441, Yinchuan, China, Aug. 2013.
[5] C. Zhou, G. Yang, D. Liang, X. Yang, and B. Xu, “An Integrated Skeleton Extraction and Pruning Method for Spatial Recognition of Maize Seedlings in MGV and UAV Remote Images,” IEEE Transactions on Geoscience and Remote Sensing Society, vol. 56, no. 8, pp. 4618-4632, Aug. 2018.
[6] A. A. A. Rahim, H. Hashim, N. E. Abdullah, S. L. M. Hassan, I. Shairah, A. Halim, and F. A. Igol, “A numerical analysis of correlation between sucrose level measurement and near-infrared (NIR) for various grades of watermelon ripeness,” 2013 International Conference on Technology, Informatics, Management, Engineering and Environment, pp.180-185, Bandung, Indonesia, June 2013.
[7] M. Phothisonothai, S. Tantisatirapong, and A. Aurasopon, “Automated determination of watermelon ripeness based on image color segmentation and rind texture analysis,” 2016 International Symposium on Intelligent Signal Processing and Communication Systems, pp.1-5, Phuket, Thailand, Oct. 2016.
[8] S. B. Kutty, N. E. Abdullah, H. Hashim, A. A. A. R, A. S. Kusim, T. N. T. Yaakub, P. N. A. M. Yunus, and M. F. A. Rahman, “Classification of Watermelon Leaf Diseases Using Neural Network Analysis,” 2013 IEEE Business Engineering and Industrial Applications Colloquium, pp.459-464, Langkawi, Malaysia, Apr. 2013.
[9] B. Liu, E. Blash, Y. Chen, D. Shen, and G. Chen, “Scalable Sentiment Classification for Big Data Analysis Using Naïve Bayes Classifier,” 2013 IEEE International Conference on Big Data, pp. 99-104, Silicon Valley, CA, USA, Oct. 2013.
[10] A. McCallum and K. Nigam, “A comparison of event models for Naïve Bayes text classification,” In Proceedings of AAAI-98, Workshop on Learning for Text Categorization, vol. 752,pp.41-48,1998.
[11] Nobuyuki Otsu, “A Threshold Selection Method from Gray-Level Histograms,” IEEE Transactions on System, Man, and Cynernetics, vol. 9, no. 1, pp.62-66, Jan. 1979.
[12] 分辨標準量測法
http://140.134.32.129/scteach/mech/text/yin01_1.html, 6, Aug. 2019.
[13] D. Hand and K. Yu, “Idiot's Bayes: Not So Stupid After All?” International Statistical Review / Revue Internationale de Statistique, Vol. 69, no. 3, pp.385-398, Dec. 2001.
[14] C. C. Chang and C. J. Lin, “LIBSVM: A library for support vector machines,” ACM Transactions on Intelligent Systems and Technology, Volume 2 Issue 3, Apr. 2011.
[15] S. L. Salzberg “C4.5: programs for machine learning,” Morgan Kaufmann Publishers Inc. San Francisco, CA, USA, 1993.
[16] R. Haralick, S. Sternberg, and X. Zhuang, “Image Analysis Using Mathematical Morphology,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 9, no. 4, pp. 532-550, July 1987.
[17] Opencv-contour函式庫
https://docs.opencv.org/master/d3/d05/tutorial_py_table_of_contents_contours.html, 6, Aug. 2019.
[18] T. Fawcett, “An introduction to ROC analysis,” Pattern Recognition Letters, vol. 27, no. 8, pp. 861-874, 2006.