English  |  正體中文  |  简体中文  |  Items with full text/Total items : 28588/40619
Visitors : 4197367      Online Users : 51
RC Version 4.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Adv. Search
LoginUploadHelpAboutAdminister

Please use this identifier to cite or link to this item: http://ntour.ntou.edu.tw:8080/ir/handle/987654321/52884

Title: 基隆港船舶影像的排煙偵測和分析
Smoke Detection and Analysis of Ship Images in Keelung Harbor
Authors: Yang, Tsung-Ju
楊宗儒
Contributors: NTOU:Department of Marine Environmental Informatics
國立臺灣海洋大學:海洋環境資訊系
Keywords: 船舶排煙;基隆港;影像分析;紅外光;自動化偵煙
Ship smoke;Keelung Harbor;Image analysis;Infrared;Automated smoke detection
Date: 2018
Issue Date: 2020-01-20T06:20:40Z
Abstract: 基隆港近年來逐漸轉型為觀光港口,且將以台灣的郵輪母港為目標進行經營,郵輪的停靠近3年來104年進出港郵輪190次、105年為222次以、106年269次,有著明顯的成長。而郵輪停靠的區域鄰近市區,為基隆港的門面,因此在船舶排煙的管制上,期望能透過影像資料的分析,進而實現偵煙自動化的目的。 船舶排煙在以往的研究上,是以可見光(紅光、綠光、藍光)資料分析為主,本研究除了可見光外,運用了紅外光熱像儀擷取影像資料,增加了溫度波段的變化,進行基隆港船舶排煙影像的偵測及分析。在研究方法上,先使用ENVI遙測影像分析軟體讀取由6個波段(R、G、B、溫度、溫度梯度及紅光梯度)組成多光譜影像資料。在監督式分析(使用Minimum Distance)結果中,屬於幾乎完全吻合(almost perfect)的共有11張影像、屬於高度吻合(substantial)的共有12張影像屬於中等吻合度(moderate)的共有9張影像、屬於一般吻合度(fair)的共有2張影像。而以非監督式,用K-mean分類法與感興趣區ROI(region of interest)進行驗證。結果顯示,有8個影像屬於高度吻合(substantial),有26張影像屬於幾乎完全吻合,但有部份影像資料會受水波紋干擾,進而降低判煙的準確度。 而在自動化條件偵煙部分,先用人工篩選出船舶排煙的特性及條件,進而透過各波段資料的特性,建立自動偵測排煙,結果顯示,此方法能明顯的偵測出黑煙且能過濾掉煙度值較灰的灰煙,本研究資料34張影像中共有12張黑煙影像及22張淡煙影像,黑煙影像資料電腦皆能準確的自動判斷出煙區,準確率達100%,淡煙影像有3張誤判黑煙,準確率為86.36%。
In recent years, Keelung Harbor has gradually transformed into a tourist port, and aims to become the Taiwan's cruise home port. The cruises entered Keelung Harbor nearly 190 times in 2015, 222 times in 2016, 269 times in 2017, respectively. The docking area of the cruises is adjacent to the downtown, where is also the portal of Keelung. Therefore, to control the smoke exhaust of the ship, it is expected to establish the automatic detection by analyzing the image data. In the previous research, the detection of ship smoke was based on the analysis of visible (red, green, and blue) bands. In this study, in addition to visible band, the thermal infrared is used to capture images and the data of temperature variation are also included for detection and analysis of ship smoke images of Keelung Harbor. Firstly, the software, ENVI is used to analyze the multi-spectral image data composed of six bands (Red, Green, Blue, Temperature, Temperature gradient and Red gradient). Through the supervised classification method, the results of agreement show that 11 images are almost perfect, 12 images are substantial, 9 images are moderate, and 2 images are fair. Through the analysis of unsupervised K-means classification method and the validation of the region of interest, the results of agreement show that 8 images are substantial, and 26 images are almost perfect, but some of the images are disturbed by striped signals, which reduce the accuracy. The result shows that the automatic detection algorithm is available to detect the black smoke and exclude the grayer smoke. Twelve images in this study are the black-smoke, and twenty-two are gray-smoke ones. The automatic detection algorithm can exactly detect the black smoke images and the accuracy rate is 100%. Nevertheless, three gray images are still misjudged as black smoke, so its accuracy is 86.36%. Keywords: Ship smoke, Keelung Harbor, Image analysis, Infrared, Automated smoke detection
URI: http://ethesys.lib.ntou.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=G004054E014.id
http://ntour.ntou.edu.tw:8080/ir/handle/987654321/52884
Appears in Collections:[海洋環境資訊系] 博碩士論文

Files in This Item:

File SizeFormat
index.html0KbHTML1View/Open


All items in NTOUR are protected by copyright, with all rights reserved.

 


著作權政策宣告: 本網站之內容為國立臺灣海洋大學所收錄之機構典藏,無償提供學術研究與公眾教育等公益性使用,請合理使用本網站之內容,以尊重著作權人之權益。
網站維護: 海大圖資處 圖書系統組
DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback