Sekundarni jezik: |
Slovenski jezik |
Sekundarni naslov: |
Koregistracija trirazsežnih modelov stavb z grafičnimi gradniki zaznanimi z infrardečih aero videoposnetkov |
Sekundarni povzetek: |
In the European Union (EU) countries buildings consume 40% of the energy and cause 36% of CO2 emissions.
The thermal information of facades and roofs are important for building inspection and energy saving. Texturing
the existing three-dimensional (3D) building models with infrared (IR) images enriches the model database and
enables analysis of energy loss of buildings.
The main purpose of the presented thesis is to investigate methods for automatic extraction of the IR textures for
roofs and facades of the existing building model. The correction of the exterior orientation parameters of the IR
camera mounted on mobile platform is studied. The developed method bases on a point-to-point matching of the
features extracted from IR images with a wire frame building model.
Firstly, extraction of different feature types is studied on a sample IR image; Förstner and intersection points are
chosen for representation of the image features. Secondly, the 3D building model is projected into each frame of
the IR video sequence using orientation parameters; only coarse exterior orientation parameters are known. Then
the automatic co-registration of a 3D building model projection into the image sequence with image features is
carried out. The matching of a model and extracted features is applied iteratively and exterior orientation
parameters are adjusted with least square adjustment. The method is tested on a dataset of dense urban area.
Finally, an evaluation of developed method is presented with five |
Sekundarne ključne besede: |
geodezija;diplomska dela;UNI;3D modeli stavb;infrardeč spekter;digitalna obdelava podob;registracija podob;videoposnetki;fotogrametrija; |
Vrsta datoteke: |
application/pdf |
Vrsta dela (COBISS): |
Diplomsko delo |
Komentar na gradivo: |
Univ. Ljubljana, Fak. za gradbeništvo in geodezijo |
Strani: |
XXX, 127 str. |
Vrsta dela (ePrints): |
thesis |
Naslov (ePrints): |
Co-registration of three-dimensional building models with image
features from infrared video sequences |
Ključne besede (ePrints): |
3D modeli stavb;infrardeč� spekter;digitalna obdelava podob;registracija podob;videoposnetki;fotogrametrija |
Ključne besede (ePrints, sekundarni jezik): |
3D building model;infrared radiation;image processing;image registration;feature extraction;photogrammetry;image sequences;matching |
Povzetek (ePrints): |
V državah Evropske unije (EU) porabijo zgradbe 40 % energije in povzročijo 36 % vseh emisij CO2. Podatki o
temperaturi fasad in streh so pomembni pri določitvi temperaturne učinkovitosti zgradb, s čimer omogočajo
prihranek energije. Teksturiranje obstoječega trirazsežnega (3D) modela stavb z infrardečimi (IR) posnetki
dopolni podatkovno bazo modela zgradb in omogoči analize njihovih energetskih izgub.
Namen raziskovalne naloge je samodejna določitev tekstur streh in fasad stavb obstoječega trirazsežnega (3D)
modela z IR posnetkov. Za to je potrebno izboljšanje natančnosti parametrov zunanje orientacije IR kamere
pritrjene na mobilno platformo. Razvita metoda temelji na ujemanju točk samodejno (avtomatsko) zaznanih
grafičnih gradnikov z IR videoposnetka in žičnega modela stavb.
V nalogi smo najprej proučili zaznavo različnih tipov grafičnih gradnikov na testnem IR posnetku. Förstnerjeve
in presečiščne točke smo izbrali kot primerne grafične gradnike za predstavitev obravnavanih značilnosti stavb
na IR posnetku. 3D model stavb je projiciran na vsak posamezen posnetek videosekvence ob upoštevanju
orientacijskih parametrov, od katerih so parametri zunanje orientacije podani le s približnimi vrednostmi. Nato
smo izvedli samodejno koregistracijo 3D modela stavb projiciranega na videoposnetek in grafičnih gradnikov
zaznanih z istega IR videoposnetka. V iterativnem postopku samodejnega ujemanja 3D modela stavb in zaznanih
grafičnih gradnikov smo parametre zunanje orientacije izravnali z metodo najmanjših kvadratov. Razvito
metodologijo za koregistracijo in izravnavo zunanjih orientacijskih parametrov smo preizkusili na strnjenem
poseljenem območju. Kvaliteto metodologije smo ocenili s petimi parametri: učinkovitostjo metodologije, ter
popolnostjo in pravilnostjo algoritmov za ujemanje in zaznavo grafičnih gradnikov. |
Povzetek (ePrints, sekundarni jezik): |
In the European Union (EU) countries buildings consume 40% of the energy and cause 36% of CO2 emissions.
The thermal information of facades and roofs are important for building inspection and energy saving. Texturing
the existing three-dimensional (3D) building models with infrared (IR) images enriches the model database and
enables analysis of energy loss of buildings.
The main purpose of the presented thesis is to investigate methods for automatic extraction of the IR textures for
roofs and facades of the existing building model. The correction of the exterior orientation parameters of the IR
camera mounted on mobile platform is studied. The developed method bases on a point-to-point matching of the
features extracted from IR images with a wire frame building model.
Firstly, extraction of different feature types is studied on a sample IR image; Förstner and intersection points are
chosen for representation of the image features. Secondly, the 3D building model is projected into each frame of
the IR video sequence using orientation parameters; only coarse exterior orientation parameters are known. Then
the automatic co-registration of a 3D building model projection into the image sequence with image features is
carried out. The matching of a model and extracted features is applied iteratively and exterior orientation
parameters are adjusted with least square adjustment. The method is tested on a dataset of dense urban area.
Finally, an evaluation of developed method is presented with five |
Ključne besede (ePrints, sekundarni jezik): |
3D building model;infrared radiation;image processing;image registration;feature extraction;photogrammetry;image sequences;matching |
ID: |
8309832 |