Segmentacija slijeda slika metodama oduzimanja pozadine

scientific master's theses

• Title (english): Image Sequence Segmentation Based on Background Subtraction Methods
• Author: Josip Krapac
• Mentor: Slobodan Ribarić (mentor)
• Committee member: Sven Lončarić (predsjednik povjerenstva)
• Committee member: Slobodan Ribarić (član povjerenstva)
• Committee member: Nikola Pavešić (član povjerenstva); VIAF: 305473372
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Electronics, Microelectronics, Computer and Intelligent Systems); Zagreb
• Defense date and country: 2006-10-30, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing; Artificial Intelligence
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing; Data Processing
• Universal decimal classification (UDC): 004 - Computer science and technology. Computing. Data processing

Abstract

Jedan od važnih zadataka niže razine u sustavima računalnog vida je segmentacija scene. Segmentacija temeljena na kretanju odvaja pokretne objekte od statičnih dijelova scene. U radu je dan pregled metoda koje segmentiraju scenu na temelju pokreta, s posebnim naglaskom na metode oduzimanja pozadine. Implementirane metode oduzimanja pozadine testirane su na umjetno generiranim i stvarnim slijedovima slika. Rezultati su usporedeni na temelju mjere kvalitete klasifikacije. Istražena je moguc´nost uporabe linearne diskriminantne analize u oblikovanju modela pozadine, te ukazano na probleme ovog pristupa. Predloženi su modeli pozadine, izgradeni prototipovi, te ocjenjena njihova uspješnost na testnim slijedovima slika. Metode oduzimanja pozadine prepostavljaju nepokretnu kameru. U sklopu rada oblikovan je model sustava za segmentaciju slijeda slika pokretne ruke uporabom kamere nagibzakret. Zajednički prikaz scene ostvaren je izgradnjom cilindrične panorame. Takav prikaz je omoguc´io uporabu metoda oduzimanja pozadine u slučaju kada je kretanje kamere ograničeno na zakret oko optičkog centra kamere.

Abstract (english)

Scene segmentation presents an important first task of many computer vision systems. Motion-based segmentation separates moving objects form static parts of the scene. An overview of motion-based segmentation methods is given, with an emphasis on background subtraction methods. Implemented background subtraction methods were tested both on artificially generated and real image sequences. Results of tested methods were compared using a measure of classifier quality. This work explored the use of linear discriminant analysis in design of background model, and pointed out problems of this approach. Proposed background models were implemented and their efficacy in segmenting test image sequences was evaluated. Background subtraction methods assume static camera. In this work a model for segmentation image sequence, acquired by pan-tilt camera, presenting movement of human hand was proposed. Common scene representation was accomplished by construction of cylindrical panorama. This representation allowed use of background subtraction methods in the moving camera case, when movement of the camera is constrained to rotation about camera’s optical center.

• Keywords: računalni vid; segmentacija temeljena na kretanju; oduzimanje pozadine; lineaerna diskriminantna analiza; cilindrična panorama
• Keywords (english): computer vision; motion-based segmentation; background subtraction; linear discriminant analysis; cylindrical panorama
• Language: croatian
• URN:NBN: urn:nbn:hr:168:394080
• Study programme: Title: Postgraduate master programe in computing; Study programme type: university; Study level: postgraduate; Academic / professional title: magistar inženjer računarstva (magistar inženjer računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-01-17 11:54:54