Detekcija kratera iz digitalnih topografskih slika

doctoral thesis

• Title (english): Crater Detection from Digital Topography Images
• Author: Goran Salamunićcar
• Mentor: Sven Lončarić (mentor)
• Committee member: Slobodan Ribarić (predsjednik povjerenstva)
• Committee member: Marko Subašić (član povjerenstva)
• Committee member: Tihomir Marjanac (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Electronic Systems and Information Processing); Zagreb
• Defense date and country: 2012-04-24, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing
• Universal decimal classification (UDC): 004 - Computer science and technology. Computing. Data processing

Abstract

U ovome radu predložena je nova metoda za detekciju kratera iz digitalnih topografskih slika. Rezultati su: (1) okruženje za razvoj i osnovno vrednovanje algoritama za detekciju kratera (CDA ova) koja uključuje mjeru udaljenosti između kratera (u području: planetarna širina, planetarna visina, dijametar); (2) katalog za vrednovanje CDA-ova s 57633 kratera katalogiziranih bez korištenja CDA ova te metodi za provjeru konzistentnosti i integraciju kataloga; (3) metoda za detekciju kratera koristeći detektor ruba i gradijent, izmijenjenu Houghovu transformaciju, morfometrijska mjerenja i analizu topografije te parametarskog prostora Houghove transformacije, podešavanje pomicanjem i kalibraciju; (4) metodologija za napredno vrednovanje CDA-ova korištenjem topografskih i optičkih slika, kontekstno temeljene metode za registraciju kratera između kataloga, interaktivne obrade kandidata za krater i registraciji u katalog; (5) poboljšanje detekcije kratera metodom interpolacije prilagođene obliku kratera, koja se pokazala učinkovita prilikom detekcije malih kratera iz topografija Marsa i Mjeseca; (6) dodatno poboljšanje detekcije kratera strojem s potpornim vektorima, čime se postiže bitno manje pogrešnih detekcija. Dodatni rezultati su najpotpuniji javno dostupni katalozi Marsa i Mjeseca s 132843 odnosno 64215 kratera te primjene za: topografiju Fobosa (katalog s 1102 kratera), vrednovanje laboratorijskih kratera, topografiju rekonstruiranu iz optičkih slika Mjeseca te topografije Venere (katalog s 345 kratera) i Zemlje (katalog s 24 kratera).

Abstract (english)

This work proposes a new method for detection of craters from digital topography images. The results are: (1) a framework for development and basic evaluation of crater detection algorithms (CDAs), which includes a measurement definition for distances between craters (in longitude-latitude-diameter space); (2) a catalogue for evaluation of CDAs, which consists of 57633 craters catalogued without usage of CDAs, including methods for evaluation of consistency and for integration of catalogues; (3) a method for crater detection using edge detection and gradient, modified Hough transform, morphometry measurements and analysis of topography and parameter space of Hough transform, slip-tuning, and calibration; (4) methodology for advanced evaluation of CDAs using topography and image datasets, including a method for registration of craters between different catalogues based on context, interactive evaluation of crater-candidates and registration in catalogue; (5) an improvement of crater detection utilizing a Crater Shape-based interpolation method that proved to be efficient for the detection of very small craters from the topographies of Mars and the Moon; and (6) an additional improvement of crater detection using support vector machines, which considerably decreases false detections. The accompanying results are the most complete publically available catalogues for Mars and the Moon, with 132843 and 64215 craters, respectively, and applicability for: Phobos’ topography (catalogue with 1102 craters), evaluation of laboratory craters, topography reconstructed from Lunar optical images, and topographies of Venus (catalogue with 345 craters) and Earth (catalogue with 24 craters).

• Keywords: detekcija kratera; detektori ruba; Houghova transformacija; stroj s potpornim vektorima
• Keywords (english): crater detection; edge detection; Hough transform; Supported Vector Machines.
• Language: croatian
• URN:NBN: urn:nbn:hr:168:205918
• Study programme: Title: Computer Science; Study programme type: university; Study level: postgraduate; Academic / professional title: Doktor znanosti (Doktor znanosti)
• Catalog URL: http://lib.fer.hr/cgi-bin/koha/opac-detail.pl?biblionumber=38854
• Type of resource: Text
• Extent: 237 str ; 30 cm
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2019-07-17 08:11:15