Publications

Detailed Information

Smoke detection algorithm in video surveillance system : 영상감시 시스템에서의 연기감지 알고리즘

DC Field Value Language
dc.contributor.advisor강명주-
dc.contributor.author고광현-
dc.date.accessioned2017-07-19T08:58:30Z-
dc.date.available2017-07-19T08:58:30Z-
dc.date.issued2013-02-
dc.identifier.other000000009491-
dc.identifier.urihttps://hdl.handle.net/10371/131463-
dc.description학위논문 (석사)-- 서울대학교 대학원 : 수리과학부, 2013. 2. 강명주.-
dc.description.abstract본 논문은 기본적인 영상처리 기법들을 이용하여 영상감시 시스템에서의 연기감지 알고리즘을 제안한다. 일반 적으로 불 또는 연기의 화재 감시를 위해서는 별도의 장치를 사용한다. 하지만, 이미 널리 이용이 되고 있는 폐쇄회로(CCTV)를 이용하면 별도의 추가적 비용 없이 화재를 감시 할 수 있다. 영상 정보만으로 연기를 감시하는 시스템은 연기의 움직임, 확산성, 색정보를 사용한다. 본 논문에서는 일반적으로 배경의 움직임이 적으며, 밝기가 어느정도 있는 실내, 야외 공간에서의 연기감지 알고리즘을 제안한다. 이를 위해, 입력영상으로부터 영상차방법과 블록화를 통하여 움직임 영역을 추출하고, 밝기정보와 색농도 정보를 이용하여 연기색 후보 영역을 추출하며, 소벨 마스킹을 통한 해당 영역의 엣지 정보를 확인한다. 그리고 확산이 확인된 경우, 방향성과 적률불변량계산을 통하여 최종적으로 연기영역인지 판단하고 연겨경보를 발령하는 시스템을 구현한다.-
dc.description.abstractIn this paper, we propose a smoke detection algorithm in video surveillance system using by the basic image processing. In general, to detect fire or smoke, a separate device is required for a surveillance system, this system, however, can by implemented by using widely used closed-circuit television(CCTV), which does not need separate devices and extra cost. The systems called video-based smoke surveillance systems use mainly a method extracting motion, diffusibility(spreading property), and chromatic information from an input image only. This paper deals with a surveillance system which is state, bright indoor and outdoor. The proposed algorithm extracts the moving object from the input image using by image difference method and block processing, the chromatic feature using by intensity and saturation space, makes a decision whether the object is smoke or not by means of the edge obtained by Sobel masking. And check a motion orientation and invariant moment of candidate regions when these regions are spread. And issues a smoke alarm when the above methods are true.-
dc.description.tableofcontents1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Previous work . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 Basic image processing 3
2.1 How to see image and video in numerics . . . . . . . . . . . . 3
2.2 Color spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.1 RGB space . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2.2 HSI space . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Image filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3 Moving object detection 12
3.1 Image difference methods . . . . . . . . . . . . . . . . . . . . . 13
3.1.1 Background modeling and subtraction . . . . . . . . . 13
3.1.2 Frame difference and clustering . . . . . . . . . . . . . 14
3.2 Block processing . . . . . . . . . . . . . . . . . . . . . . . . . 15

4 Chromatic features detection 19
4.1 RGB similarity . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2 Intensity range . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.3 Saturation range . . . . . . . . . . . . . . . . . . . . . . . . . 21

5 Edge and blurred area detection 24
5.1 First-order derivative and gradient . . . . . . . . . . . . . . . 24
5.2 Derivation of Sobel mask . . . . . . . . . . . . . . . . . . . . . 26
5.3 Blurred area detection method . . . . . . . . . . . . . . . . . . 27

6 Dynamic features detection 30
6.1 Motion orientation estimation . . . . . . . . . . . . . . . . . . 30
6.2 Invariant moment feature analysis . . . . . . . . . . . . . . . . 32

7 Experimental results 38
7.1 Suggested algorithm flow . . . . . . . . . . . . . . . . . . . . . 38
7.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
-
dc.formatapplication/pdf-
dc.format.extent2127583 bytes-
dc.format.mediumapplication/pdf-
dc.language.isoen-
dc.publisher서울대학교 대학원-
dc.subjectImage processing-
dc.subjectComputer vision-
dc.subjectSmoke detection-
dc.subjectVideo surveillance-
dc.subject.ddc510-
dc.titleSmoke detection algorithm in video surveillance system-
dc.title.alternative영상감시 시스템에서의 연기감지 알고리즘-
dc.typeThesis-
dc.contributor.AlternativeAuthorKwanghyun Ko-
dc.description.degreeMaster-
dc.citation.pagesv, 44-
dc.contributor.affiliation자연과학대학 수리과학부-
dc.date.awarded2013-02-
Appears in Collections:
Files in This Item:

Altmetrics

Item View & Download Count

  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Share