Large-scale, country-wide VHR photogrammetric 3D mapping
This is a modal window.
Das Video konnte nicht geladen werden, da entweder ein Server- oder Netzwerkfehler auftrat oder das Format nicht unterstützt wird.
Formale Metadaten
| Titel |
| |
| Serientitel | ||
| Anzahl der Teile | 53 | |
| Autor | ||
| Lizenz | CC-Namensnennung 3.0 Deutschland: Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen. | |
| Identifikatoren | 10.5446/43940 (DOI) | |
| Herausgeber | ||
| Erscheinungsjahr | ||
| Sprache |
Inhaltliche Metadaten
| Fachgebiet | |
| Genre |
FOSS4G Europe 201841 / 53
12
17
22
23
30
38
47
49
00:00
Prozess <Informatik>Digitale SignalverarbeitungProzess <Informatik>StellenringDienst <Informatik>MultiplikationsoperatorMAPp-BlockDistributionenraumCASE <Informatik>GoogolMapping <Computergraphik>SpeicherabzugDatenverwaltungZahlenbereichTaskDatenerfassungComputeranimationVorlesung/Konferenz
01:30
Digitale SignalverarbeitungProzess <Informatik>Keller <Informatik>SoftwarePunktSkalierbarkeitDatenverwaltungEin-AusgabeRechenschieberRoutingMAPResultanteQuadratzahlQuaderGrenzschichtablösungSchreib-Lese-KopfProgrammbibliothekPunktAbgeschlossene MengeDigitale PhotographieFlächentheorieClientCoxeter-GruppeGamecontrollerFunktion <Mathematik>Dienst <Informatik>SoftwareentwicklerBitrateDatenverwaltungGenerator <Informatik>Open SourceSchnittmengePhysikalisches SystemProjektive EbeneSplineSoftwareProdukt <Mathematik>Überlagerung <Mathematik>MereologieProzess <Informatik>ComputerspielSkalierbarkeitPixelp-BlockInformationEndliche ModelltheorieSatellitensystemDifferenteDivisionKontextbezogenes SystemFlächeninhaltInformationsspeicherungTesselationHydrostatikVerkehrsinformationDiagrammPolygonnetzRegulärer GraphVektorraumDreieckMultiplikationDynamisches SystemBandmatrixServerBenutzerbeteiligungKartesische KoordinatenSystemplattformIntegralStreaming <Kommunikationstechnik>Vorlesung/Konferenz
07:08
DistributionenraumQuellcodeDemo <Programm>GruppenoperationMAPInhalt <Mathematik>Mathematische LogikOrdnungsreduktionKartesische KoordinatenLastDigitale PhotographieAuflösung <Mathematik>Produkt <Mathematik>FlächentheorieTextur-MappingMultiplikationsoperatorDatenkompressionEndliche ModelltheorieKontextbezogenes SystemPolygonnetzSichtenkonzeptStreaming <Kommunikationstechnik>SatellitensystemSchlussregelKlassische PhysikMereologieServerSchnittmengeCoxeter-GruppeProzess <Informatik>Mapping <Computergraphik>RandwertSchlüsselverwaltungGlobale OptimierungDifferenteRechenwerkSpieltheorieRadiusEnergiedichteHilfesystemKonfigurationsraumGewicht <Ausgleichsrechnung>VerschlingungProjektive EbeneTermBildverstehenAbgeschlossene MengeRuhmasseDistributionenraumDienst <Informatik>SprachsyntheseVorlesung/Konferenz
Transkript: Englisch(automatisch erzeugt)
00:18
CZ is a local market leader ever since,
00:23
because CZNAM.CZ, it's one of those companies who is crazy enough to compete with Google on local market. And now we get to the core of this talk, how we created the VHR 3D map of Czech Republic.
00:44
The mapping process can be divided to four sub-tasks. Data acquisition, data processing, data management and fusion, and data distribution.
01:01
So firstly, briefly about data acquisition. Data acquisition is handled for us by third party partner and data are delivered in blocks in various time and it's a continuous process.
01:23
So you can see on this map, the whole country is divided into blocks. And I'm sorry, yeah, we are back. So the whole country is divided into blocks and the darker part, it's areas with better GSD pictures,
01:51
imagery. OK, let's see slightly into processing. The processing is done with our fully automatic
02:03
photogrammetric system. System creates all output data sets in a single unified process. So we have a digital surface model and auto photo and true 3D map on output. The only manual input is a definition
02:23
of ground control points of which we needed a lot. The process consists of input quality control, GCP definition, photogrammetric system initialization, processing, output quality control, and packaging for delivery.
02:44
Of course, we didn't process all area together. We have split the whole Czech Republic to 350 square parts called targets. And one target has about 225 square kilometers.
03:06
The targets were processed and delivered gradually. And at this diagram on slide, you can see the moped area division with the status
03:22
report information of each target. But the important part I want to talk about are the last two steps. These steps we have covered with our open source VTS 3D
03:43
geospatial software stack, which is the second project I talked about in the beginning of this talk. And let's talk about VTS briefly. VTS is an integrated platform for 3D map application
04:04
development. And it's a virtual landscape streaming and rendering engine. And as I mentioned before, it's fully open sourced and BSD to close license.
04:21
There was more talks about VTS yesterday. And today will be another two just after this presentation. So feel welcome to stay and listen my colleague to talk about VTS closely.
04:42
The strongest point of VTS are data scalability. Oh, sorry. Data scalability, you can fuse together countless amount of 3D and 2 and 1.5D data. High performance streaming servers
05:01
with bandwidth optimized dynamic triangle regular mesh. Auto photo generation and static tile streaming. And VTS provides lightweight and fast client libraries
05:20
for web and desktop. OK, now let's talk about the data management. At the end, we are handling more than 10 terabytes of data.
05:49
So and the result map consists of many data sets. So there is a global context, which is a global dam with a satellite imagery.
06:01
There are multiple true 3D models with various GSD. As I mentioned before, there are extra VLAN areas with 12.5 centimeter per pixel, and major cities with 10 centimeter per pixel.
06:22
And of course, there are vector data with labels and features. And VTS is able to fuse those data sets together very, very smoothly. And the process of Fusion must be reentrant and, of course,
06:45
low cost, because it's happening a lot. More about Fusion, you will hear from my colleague Ladislav in the following talk. So let's move on.
07:03
As I mentioned before, the data from photogrammetric software comes in targets. And on production servers, our data joined to logical groups, like cities, landmarks,
07:21
or extra VLAN. And in a VTS map configuration, are those logical groups referenced as one data set. It's very helpful for quality of streaming and rendering the resulting map.
07:46
And the last part of mapping process is distribution. To provide the best user experience which is desired,
08:01
we need to have in mind basic rules of what is important for streaming. And streaming needs as few data as possible. So in VTS, we optimize the streaming by advanced mesh compression and by texture packing.
08:29
The user experience is crucial to speed up the first load time. So VTS help us there with virtual surfaces.
08:45
It reduce amount of loaded metatiles and so on. And, of course, the loading policies needs to be also optimized.
09:02
In production, there is a robust content delivery network, which allows us to keep more often used data as close to user as possible. So there are caching CDN energies, and the requests on the VTS streaming server
09:25
are passed in very low frequency. The same. OK, so that's basically all about VTS and VHR 3D
09:47
photogrammetric map of Czech Republic. And please let me show you the brief tour around the portal of map it has at.
10:02
So in here, you can see the portal. It's a classical map application. A map application consists of a whole world. There is an auto photo satellite imagery. And in Czech Republic, we have better data.
10:28
But what is important in this talk is this 3D view button. When it's activated, the portal is switched to 3D view.
10:43
So you can see that it's no more projected map, but it's a globe with a whole world context. But again, in Czech Republic, you can see that there are better data.
11:00
There are those 3D model of whole Czech Republic. And we are zooming to some major city to see this in action.
11:29
Here, you can see the high resolution, VHR resolution map of Prague. And the last thing I want to show you
11:42
is the nice and smooth boundary between two very different data sets. And it's staying, which will be covered with my colleague Gladys Lavorky in the next talk.
12:02
So that's all from me. And now we may open the floor for questions. Thank you very much, Thomas. Nice presentation. Thank you. Do you have any question about this presentation now?
12:20
It's clear? OK, so we finish here this presentation. And we are going to thank you very much again. Thank you. And we are going to wait about one, two minutes if anybody wants to come from the other room.