TILEN URBANČIČ, DOKTOR ZNANOSTI Dne 24. novembra 2017 je v okviru doktorskega študija geodezije na Fakulteti za gradbeništvo in geodezijo Univerze v Ljubljani (UL FGG) doktorsko nalogo uspešno ...zagovarjal Tilen Urbančič, univ. dipl. inž. geod., ki je nalogo pripravil pod mentorstvom prof. dr. Bojana Stoparja in doc. dr. Mojce Kosmatin Fras, obeh s Fakultete za gradbeništvo in geodezijo pri Univerzi v Ljubljani. Assessment of Geometric Similarity of Airborne Laser Scanning Point Clouds) Mentorica: prof. dr. Bojan Stopar Somentorica: doc. dr. Mojca Kosmatin Fras URL: https://repozitorij.uni-lj.si/IzpisGradiva.php?id=99245&lang=slv Doktorska naloga se ukvarja z ocenjevanjem geometrične podobnosti oblakov točk, kjer pri registraciji uporabimo umetne tarče. Utilization of spatial data sets and their cartographic visualization for the purpose of real properties taxation) Mentor: doc. dr. Dušan Petrovič Somentorica: prof. dr. Andreja Cirman URL: https://repozitorij.uni-lj.si/IzpisGradiva.php?id=99246&lang=slv V doktorski disertaciji so obravnavani prostorski podatki, povezani z obdavčitvijo nepremičnin, in njihovi kartografski prikazi, namenjeni vsem v proces obdavčitve nepremični vključenim deležnikom, odlaične javnosti do pripravljavcev politik in strokovnih podlag ter odločevalcev.
The Sun is the main energy source of the life on the Earth. Thus, solar radiation energy data and models are important for many areas of research and applications. Many parameters influence the ...amount of solar energy at a particular standing point of the Earth's surface; therefore, many solar radiation models were produced in the last few years. Solar radiation energy depends mostly on incidence angle, which is defined by astronomical and surface parameters. Our solar radiation model is based on defining incidence angle by computing normal-to-the-surface tangent plane and direction of the Sun. If a part of the surface is in the shadow, it receives lesser energy than sunny areas. That is why shadow determination is an important part of the model. The sky is usually not completely clear, so meteorological parameters had to be integrated into the model. Meteorological model distinguishes among direct and diffuse Sun radiation. The model was tested and implemented for the whole Slovenia and it was also compared with previous studies. Case study surface data were calculated from the DEM with a 25
m resolution. The astronomical data, which were required for virtual Sun motion simulation around the Earth, were derived from the astronomical almanac. Meteorological data were acquired from observed mean values on 24 meteorological stations between 1961 and 1990. All calculations were made for hours and decades and finally, the annual quasiglobal radiation energy, which is the energy received by inclined plane from the Sun in one year, was calculated from the sum of all the energies of all the decades.
Remote sensing has become the most important data source for the digital elevation model (DEM) generation. DEM analyses can be applied in various fields and many of them require appropriate DEM ...visualization support. Analytical hill-shading is the most frequently used relief visualization technique. Although widely accepted, this method has two major drawbacks: identifying details in deep shades and inability to properly represent linear features lying parallel to the light beam. Several authors have tried to overcome these limitations by changing the position of the light source or by filtering. This paper proposes a new relief visualization technique based on diffuse, rather than direct, illumination. It utilizes the sky-view factor—a parameter corresponding to the portion of visible sky limited by relief. Sky-view factor can be used as a general relief visualization technique to show relief characteristics. In particular, we show that this visualization is a very useful tool in archaeology as it improves the recognition of small scale features from high resolution DEMs.
Rezultati klasifikacije stavb na orto fotu se uporabljajo kot vir za vzdrževanje katastra stavb. V zadnjih letih se za klasifikacijo stavb v svetu vse bolj uveljavljajo metode globokega učenja z ...uporabo konvolucijskih nevronskih mrež. V raziskavi predstavimo primer samodejne klasifikacije stavb z uporabo lastnih podatkovnih zbirk, izdelanih iz barvnih bližnje infrardečih ortofotov (BIR-R-G) in barvnih ortofotov (R-G-B). Preizkusili smo detekcijo stavb z uporabo predučenih uteži podatkovnih zbirk Microsoft Common Objects in Context (MS COCO) in ImageNet. Za detekcijo stavb smo uporabili Mask Region Convolutional Neural Network (Mask R-CNN). Namen raziskave jepreizkusiti uporabniško vrednost globokega učenja za detekcijo stavb z uporabo predučenih uteži na podatkih drugega barvnega prostora s ciljem izgradnje klasifikacijskega modela brez ponovnega učenja.
Primeri takih tekmovanj so DeepGlobe Buildings Extraction Challange7, SpaceNet Building Extraction Challenge8, crowdAI Mapping Challenge9 idr. Gre za globoko konvolucijsko nevronsko mrežo, ki se ...uporablja za detekcijo objektov (angl. object detection), semantično segmentacijo (angl. semantic segmentation) ter segmentacijo primerov (angl. instance segmentation). Massachusetts Buildings Dataset10 (Mnih, 2013), Inria Aerial Image Labeling Dataset11 (Maggiori et al., 2017), AIRS Automatic Mapping of Buildings Dataset12 (Chen et al., 2019). Klasifikacija stavb se izvaja z metodami strojnega učenja, z objektno klasifikacijo, in sicer z uporabo podpornih vektorjev (angl. support-vector machines) in naključnih gozdov (angl. random forest), kjer je ključni podatek tudi digitalni model površja.