跳到主要內容

臺灣博碩士論文加值系統

(44.223.39.67) 您好!臺灣時間:2024/05/22 15:43
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

我願授權國圖
: 
twitterline
研究生:鄒學緯
研究生(外文):ZOU,XUE-WEI
論文名稱:即時自動化生咖啡檢測系統開發基於RGB-IR多光譜影像
論文名稱(外文):Development of Real-Time Automatic Green Coffee Inspection System Using RGB-IR Multispectral Imaging
指導教授:陳士煜
指導教授(外文):CHEN, SHIH-YU
口試委員:林金樹林義隆
口試委員(外文):LIN, CHIN-SULIN, YIH-LON
口試日期:2022-07-26
學位類別:碩士
校院名稱:國立雲林科技大學
系所名稱:資訊工程系
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2022
畢業學年度:110
語文別:中文
論文頁數:70
中文關鍵詞:高光譜影像生咖啡豆深度學習高光譜生咖啡豆瑕疵檢測機台
外文關鍵詞:Hyperspectral ImageGreen Coffee BeansDeep LearningReal-time Green Coffee Bean Defect Detection Machine
數位影音連結:即時生咖啡豆瑕疵檢測機台 - 實體Demo (Real-time green coffee bean defect detection machine)
相關次數:
  • 被引用被引用:0
  • 點閱點閱:99
  • 評分評分:
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:0
咖啡豆(Coffee bean)是全球交易量僅次於石油的大宗物資,是國際貿易中不可或缺的農產品之一。咖啡豆也非常多種瑕疵,例如貝殼豆、破裂豆、全黑/局部黑豆、萎凋豆以及黴菌破壞豆等。現在台灣大部分咖啡農是用人的眼睛進行咖啡豆的篩檢,會因為視覺疲勞辨識率會隨時間增加而降低。為了解決此問題,本論文使用快照式高光譜相機(RGB-IR)取得高光譜的影像後搭配深度學習進行分類,深度學習分別為1D-CNN、2D-CNN與3D-CNN,使用混淆矩陣與kappa進行模型評比,並與其他論文進行比較,最後有與瑕疵檢測機台進行結合,打造出及時且準確率高的即時高光譜生咖啡豆瑕疵檢測機台。


Coffee beans are the world's largest commodity traded after oil, and one of the indispensable agricultural products in international trade. Coffee beans also have a wide variety of imperfections, such as shell beans, cracked beans, all black/partial black beans, withered beans, and mold-damaged beans. At present, most coffee farmers in Taiwan use human eyes to screen coffee beans, and the recognition rate will decrease with time due to visual fatigue. In order to solve this problem, this study uses a snapshot hyperspectral camera (RGB-IR) to obtain hyperspectral images and then performs classification with deep learning. The deep learning is 1D-CNN, 2D-CNN and 3D-CNN respectively, using confusion matrix and kappa The model is compared and compared with other papers. Finally, it’s combined with the defect detection machine to create a timely and accurate real-time hyperspectral green coffee bean defect detection machine.
摘要 i
ABSTRACT ii
誌謝 iii
目錄 iv
表目錄 viii
圖目錄 x
第一章 緒論 1
1.1 研究動機 1
1.2 現有問題及解決方案 1
1.3 相關研究 2
1.3.1 高光譜訊號 2
1.3.2 相關論文 4
1.4 論文架構 8
第二章 相關理論 9
2.1 深度學習(Deep Learning, DL) 9
2.2.1 卷積層(Convolution Layer) 10
2.2.1.1 標準卷積(Standard Convolution) 10
2.2.1.2 深度可分離卷積 (Depthwise Separable Convolution, SDC) 11
2.2.2 池化層(Pooling Layer) 12
2.2.2.1 最大池化(Maximum Pooling) 與 平均池化(Average Pooling) 12
2.2.2.2 全域平均池化(Global Average Pooling, GAP) 14
2.2.3 激活函數(Activation Function) 14
2.2.4 全連接層(Fully Connected Layer, FC) 16
2.2.5 Group Normalization 17
2.2 支持向量機(Support Vector Machine, SVM) 17
2.3 咖啡豆品質 18
2.4 鹵素燈與LED燈差異 22
第三章 研究方法 23
3.1 影像前處理 23
3.2 深度學習研究流程 25
3.2.1 2D-CNN 25
3.2.2 3D-CNN 27
3.3 評比標準 28
3.2.1 混淆矩陣 28
3.2.2 Cohen’s kappa 30
第四章 實驗結果與分析 32
4.1 快照式高光譜成像系統 32
4.2 咖啡豆資料蒐集 33
4.3 實驗硬體配置與參數設定 36
4.4 2D-CNN分類結果 37
4.5 3D-CNN分類結果 41
4.6 2D-CNN與3D-CNN比較結果 44
4.7 鹵素燈與LED燈比較 45
4.7.1 鹵素燈與LED燈資料蒐集 45
4.7.2 鹵素燈與LED燈比較結果 46
4.8 綜合比較結果 47
4.9 比較過往研究 50
4.10 高光譜咖啡豆即時自動化瑕疵檢測機台 51
第五章 結論 54
參考文獻 55

[1]Paolo Oliveri ; Cristina Malegori ; Monica Casale ; Edoardo Tartacca ; Gianni Salvatori. An innovative multivariate strategy for HSI-NIR images to automatically detect defects in green coffee. Journal Elsevier Talanta, July 2019, Vol:199, 270-276.
[2]Nicola Caporaso ; Martin B. Whitworth ; Stephen Grebby ; Ian D. Fisk. Rapid prediction of single green coffee bean moisture and lipid content by hyperspectral imaging. Journal Elsevier Journal of Food Engineering, July 2018, ,Vol:227,18-29.
[3]Chu Zhang ; Fei Liu ; Yong He. Identification of coffee bean varieties using hyperspectral imaging: influence of preprocessing methods and pixel-wise spectra analysis. SCIENTIFIC REPORTS, 01 Feb, 2018.
[4]Mauricio García; John E. Candelo-Becerra ; Fredy E. Hoyos. Quality and Defect Inspection of Green Coffee Beans Using a Computer Vision System. MDPI applied sciences, 2019, 9(19),4195.
[5]Edwin R. Arboleda; Arnel C. Fajardo; Ruji P. Medina.An image processing technique for coffee black beans identification. IEEE ICIRD, May 2018.
[6]N. Okubo and Y. Kurata, "Nondestructive classification analysis of green coffee beans by using near-infrared spectroscopy," Foods, vol. 8, no. 2, p. 82, 2019.
[7]B. Chu, K. Yu, Y. Zhao, and Y. He, "Development of noninvasive classification methods for different roasting degrees of coffee beans using hyperspectral imaging," Sensors, vol. 18, no. 4, p. 1259, 2018.
[8]C. Nansen, K. Singh, A. Mian, B. J. Allison, and C. W. Simmons, "Using hyperspectral imaging to characterize consistency of coffee brands and their respective roasting classes," Journal of Food Engineering, vol. 190, pp. 34-39, 2016.
[9]R. Calvini, A. Ulrici, and J. M. Amigo, "Practical comparison of sparse methods for classification of Arabica and Robusta coffee species using near infrared hyperspectral imaging," Chemometrics and Intelligent Laboratory Systems, vol. 146, pp. 503-511, 2015.
[10]J.-S. Cho, H.-J. Bae, B.-K. Cho, and K.-D. Moon, "Qualitative properties of roasting defect beans and development of its classification methods by hyperspectral imaging technology," Food chemistry, vol. 220, pp. 505-509, 2017.
[11]Wang, P., Tseng, H., Chen, T., & Hsia, C. (2021). Deep convolutional neural network for coffee bean inspection. Sensors and Materials, 33(7), 2299-2310.
[12]S.-Y. Chen, C.-Y. Chang, C.-S. Ou, and C.-T. Lien, "Detection of insect damage in green coffee beans using VIS-NIR hyperspectral imaging," Remote Sensing, vol. 12, no. 15, p. 2348, 2020.
[13]Chen, Shih-Yu, Ming-Feng Chiu, and Xue-Wei Zou. "Real-time defect inspection of green coffee beans using NIR snapshot hyperspectral imaging." Computers and Electronics in Agriculture 197 (2022): 106970.
[14]G. E. Hinton, S. Osindero, and Y.-W. Teh, “A Fast Learning Algorithm for Deep Belief Nets,” Neural Computation, vol. 18, no. 7, pp. 1527–1554, Jul. 2006, doi: 10.1162/neco.2006.18.7.1527.
[15]D. H. Hubel, T. N. Wiesel, “Receptive Fields, Binocular Interaction and Functional Architecture in the Cat’s Visual Cortex”, Journal of Physiology, pp. 106-154, 1962.
[16]Chollet, François. "Xception: Deep learning with depthwise separable convolutions." Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.
[17]M. Lin, Q. Chen, and S. Yan, “Network In Network,” arXiv:1312.4400 [cs], Mar. 2014, Accessed: May 09, 2021. [Online]. Available: http://arxiv.org/abs/1312.4400.
[18]Xavier Glorot, Yoshua Bengio, “Understanding the difficulty of training deep feedforward neural networks”, Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, pp. 249-256, 2010.
[19]K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” Proc. ICLR, 2015.
[20]Ba J L, Kiros J R, Hinton G E. Layer normalization[J]. arXiv preprint arXiv:1607.06450, 2016.
[21]Vedaldi V L D U A. Instance Normalization: The Missing Ingredient for Fast Stylization[J]. arXiv preprint arXiv:1607.08022, 2016.
[22]Joackim Mutua, POST HARVEST HANDLING AND PROCESSING OF COFFEE IN AFRICAN COUNTRIES,2000
[23]Wu, Kesheng, Ekow Otoo, and Arie Shoshani. "Optimizing connected component labeling algorithms." Medical Imaging 2005: Image Processing. Vol. 5747. SPIE, 2005.
[24]Fiorio, Christophe. "A topologically consistent representation for image analysis: the frontiers topological graph." International Conference on Discrete Geometry for Computer Imagery. Springer, Berlin, Heidelberg, 1996.
[25]Yu, C., Li, F., Chang, C.-I., Cen, K., & Zhao, M. (2018). Deep 2D convolutional neural network with deconvolution layer for hyperspectral image classification. International Conference in Communications, Signal Processing, and Systems
[26]Ahmad, M., Shabbir, S., Raza, R. A., Mazzara, M., Distefano, S., & Khan, A. M. (2021). Hyperspectral image classification: artifacts of dimension reduction on hybrid CNN. arXiv preprint arXiv:2101.10532.
[27]Paul, A., Bhoumik, S., & Chaki, N. (2021). SSNET: An improved deep hybrid network for hyperspectral image classification. Neural Computing and Applications, 33(5), 1575-1585.
[28]Chen, C., Zhang, J.-J., Zheng, C.-H., Yan, Q., & Xun, L.-N. (2018). Classification of hyperspectral data using a multi-channel convolutional neural network. International Conference on Intelligent Computing.
[29]Rao, C., & Liu, Y. (2020). Three-dimensional convolutional neural network (3D-CNN) for heterogeneous material homogenization. Computational Materials Science, 184, 109850.
[30]Mei, S., Yuan, X., Ji, J., Zhang, Y., Wan, S., & Du, Q. (2017). Hyperspectral image spatial super-resolution via 3D full convolutional neural network. Remote Sensing, 9(11), 1139.
[31]Li, Y., Zhang, H., & Shen, Q. (2017). Spectral–spatial classification of hyperspectral imagery with 3D convolutional neural network. Remote Sensing, 9(1), 67.
[32]Nagasubramanian, K., Jones, S., Singh, A. K., Sarkar, S., Singh, A., & Ganapathysubramanian, B. (2019). Plant disease identification using explainable 3D deep learning on hyperspectral images. Plant Methods, 15(1), 1-10.
[33]Yang, J., Zhao, Y.-Q., Chan, J. C.-W., & Xiao, L. (2019). A multi-scale wavelet 3D-CNN for hyperspectral image super-resolution. Remote Sensing, 11(13), 1557.
[34]Nagasubramanian, K., Jones, S., Singh, A. K., Singh, A., Ganapathysubramanian, B., & Sarkar, S. (2018). Explaining hyperspectral imaging based plant disease identification: 3D CNN and saliency maps. arXiv preprint arXiv:1804.08831.
[35]Townsend, James T. "Theoretical analysis of an alphabetic confusion matrix." Perception & Psychophysics 9.1 (1971): 40-50.
[36]McHugh, Mary L. "Interrater reliability: the kappa statistic." Biochemia medica 22.3 (2012): 276-282.

電子全文 電子全文(網際網路公開日期:20250726)
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top
無相關期刊