English  |  正體中文  |  简体中文  |  Items with full text/Total items : 27308/39152
Visitors : 2468510      Online Users : 282
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/987654321/30640

Title: 無縫感知之視訊安全偵測、隱私保護與事件搜尋整合系統-光線多變環境下之事件多樣性分析與其應用
Event Multiplicity Analysis and Its Applications under Various Lighting Conditions
Authors: 謝君偉;李建德
Contributors: NTOU:Department of Computer Science and Engineering
國立臺灣海洋大學:資訊工程學系
Keywords: 事件多樣性;抽菸事件偵測;手持物件偵測
Event multiplicity;smoking event analysis;hand-held object detection
Date: 2011-08
Issue Date: 2012-04-13T01:13:40Z
Publisher: 行政院國家科學委員會
Abstract: 摘要:所謂智慧型視覺監控系统就是利用電腦視覺的方法,在不需要人力干預的情况下,對攝影機拍 攝的影像進行自動分析,進而對目標物進行追蹤、識别和事件分析,目前大部分的事件分析技術, 都需要利用背景相減,將前景物擷取來,接著再對此前景物作特徵分析,無論如何對於戶外的環 境,背景相減法常因天候、陽光、颳風、照相機震動的影響,造成前景物偵測的失敗,例如當相 機有震動時,原來靜止的物件會被當成移動物偵測出來;當陽光移動時,樹的陰影會造成原來為 背景的路面被當成前景物偵測出來;燈的照射也會響到背景相減的結果,特別是對國內鐵路軌道 的監控環境下,特別有這種問題,例如平交道或月台,背景常因車輛或火車的車燈的照射,造成 誤判,在本子計畫中,我們將利用空間與時間軸的變化特徵點(spatial-temporal interest points),搭 配機器學習(machine learning)的技巧來分析電影、監控視訊內的異常事件,如抽菸、闖平交道等等 的事件,此方法不需建立背景,因此可以允許在照相機移動的情況下,即時偵測、分析與檢索異 常事件。此子計畫共分三年,第一年主要發展出不須背景建立的視訊特徵,在光影變化與相機移 動的情況下,從視訊中自動擷取有用的事件特徵;第二年針對事件的多樣性(multiplicity)與非靜止 相機情況下,分析各類可能發生的事件,並進行檢索;第三年針對鐵路軌道環境做各類的異常事 件預防分析,目前國內鐵路平交道因人、車不遵守規則,致使事故發生頻繁,如能事先偵測預防, 將可為鐵路行車安全帶來莫大幫助。
abstract:An intelligent surveillance system takes advantages of the technologies of computer vision to detect and track targets, and then analyze abnormal events without any human efforts. However, most current surveillance systems extract foreground objects though background subtraction which is quite unstable under different lighting and weather conditions. When the sun moves or the camera has vibrations, this subtraction technique will fail to detect foreground objects. Especially when a railroad crossing is observed, the train light or car light will lead to many background objects to be extracted through the subtraction technique. This three-year project plans to develop a robust and flexible surveillance system which detects, track, and recognize various suspicious events from videos or movies even though cameras with larger movements. In the first year, this sub-project will develop a novel feature extractor for extracting spatial-time interest features for event analysis event under various lighting conditions. In the second year, this project proposes a novel Bayesian event analyzer to deal with the problem of event multiplicity so that different abnormal events can be analyzed and searched from videos. In t he last year, all the above techniques will be integrated together for monitoring various abnormal events from railroad crossings. It sincerely hopes the integrated railroad surveillance system can prevent unexpected dangerous accidents.
Relation: NSC100-2221-E019-043-MY3
URI: http://ntour.ntou.edu.tw/handle/987654321/30640
Appears in Collections:[資訊工程學系] 研究計畫

Files in This Item:

There are no files associated with this item.



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