master's thesis
A method for forest fire detection by processing the video data

Antun Crnarić (2015)
Metadata
TitleMetoda analize video zapisa za detekciju dima
AuthorAntun Crnarić
Mentor(s)Irena Galić (thesis advisor)
Abstract
Predložena je metoda za detekciju šumskih požara analizom niza slika dobivenih iz fiksno postavljene video kamere koja snima okolinu. Prvi korak je detekcija pomaka u slici tehnikom oduzimanja pozadine primjenom modela Gaussovih mješavina. Dobivena područja se opisuju konturama nad kojima se provodi filtracija radi smanjivanja šumova i spajanje bliskih kontura kako bi se naglasile bitne informacije. Praćenjem položaja kontura kroz niz slika izdvajaju se one konture koje zadržavaju konstantan položaj. Na kraju se provodi analiza slika u RGB i YCbCr sustavu boja i donosi konačna odluka radi li se o dimu. Uspješnost rada algoritma je zadovoljavajuća.
Keywordsfire detection background subtraction contour processing smoke detection and RGB and YCbCr color space
Parallel title (English)A method for forest fire detection by processing the video data
Committee MembersIrena Galić (committee chairperson)
Časlav Livada (committee member)
Josip Job (committee member)
GranterSveučilište Josipa Jurja Strossmayera u Osijeku
Fakultet elektrotehnike, računarstva i informacijskih tehnologija Osijek
Lower level organizational unitsZavod za programsko inženjerstvo
Katedra za vizualno računarstvo
PlaceOsijek
StateCroatia
Scientific field, discipline, subdisciplineTECHNICAL SCIENCES
Electrical Engineering
Telecommunications and Informatics
Study programme typeuniversity
Study levelgraduate
Study programmeGraduate University Study Programme in Computer Engineering
Academic title abbreviationmag.ing.comp.
Genremaster's thesis
Language Croatian
Defense date2015-10-01
Parallel abstract (English)
This paper proposes a method for forest fire detection by processing the video data generated by an ordinary camera monitoring the scene. First step is background subtraction using Gaussian mixture model. For moving regions we extract contours which are then filtered by size to remove the noise and merged if they are close to emphasize more important regions. Keeping track of contour history allows us to separate contours that don't change their position on image. The decision is made after performing image analysis in RGB and YCbCr color space. Algorithm shows satisfactory results.
Parallel keywords (Croatian)detekcija požara oduzimanje pozadine obrada kontura RGB i YCbCr sustav boja detekcija dima
Resource typetext
Access conditionOpen access
Terms of usehttp://rightsstatements.org/vocab/InC/1.0/
URN:NBNhttps://urn.nsk.hr/urn:nbn:hr:200:606179
CommitterAnka Ovničević