GeoDec News:

Microsoft Research Press Release on ProDA
GeoDec in Computer Power User
GeoDec featured on Ramesh Jain's Blog
IMSC Awarded Microsoft Grant to Advance Digital Geographics
GeoDec in IMSC's 2006 Newsletter

 

"GeoDec"

Constructing an information-rich and realistic 3-dimensional visualization

 and simulation of a geographical location, rapidly and accurately.

 

GeoDec (Geospatial Decision Making) is a collaborative project with Dr. Cyrus Shahabi, Dr. Craig Knoblock , Dr. Ulrich Neumann and Dr. Ramakant Nevatiaunder IMSC to build an information-rich and realistic geospatial space (e.g., a city)  with temporal dimension rapidly and accurately, which supports visualization, querying and data analysis capabilities. Recent growth of the geo-spatial information and their availability has motivated the effort to integrate them to support a comprehensive set of queries in different modalities. However due to the inherent difference in data formats available and their different accuracy levels, doing a seamless, consistent and efficient integration of different data sources is a very challenging task.

 

By utilizing various information integration approaches such as orthoimagery and street maps conflation, vector data and satellite imagery conflation and road network and map fusion, we strive to create intelligent, information-rich and detailed models that incorporate the visual appeal and accuracy of imagery with detailed attribution information in diverse maps and realistic 3-D visualization for geographic locations.

 

 

 

This interdisciplinary work is being performed by researches from various fields such as databases, artificial intelligence, computer graphics and computer vision in a collaborative effort as part of IMSC, the only NSF engineering research center in the area of Multimedia for integration of various technologies under the GeoDec framework.

 

Our proposed framework is composed of the following components :

 

Rapid 3-D model construction from photographs:

This system is designed to minimize users' interaction by giving them the option of working with a single image at a time. At the same time, by using basic information such as one side and one corner of objects identified by user in a semi-automatic process, highly complex polygonal roof shaped buildings can be modeled.

 

Here is 3D models of USC campus with photograph texture rendered in VRML:

 

The 3D models of Downtown Los Angeles

 

The 3D models of ShangHai Jiao Tong University, China

 

 

Texture mapping of buildings and video fusion:

In a two fold approach, static and dynamics of scenes are captured from a photographic process involving video or numerous still images. Static textures capture the color attributes of buildings and street scenes and then enhanced with video streams to show dynamic aspects of scenes (e.g. movement of cars on other objects mapped on the 3-D model (video fusion).

 

To see the video demo for the following example, please click here.


 

 
 
Geospatial Data Fusion

The next step is to enhance the model by integrating all sorts of data from both online sources (e.g., yellow pages, property-tax sites, geocoders) as well as private databases (e.g., road vector data, maps, gazetteer points).

 

Integrating Vector Data and Imagery:
This research is on the problem of accurate integration of geospatial vector data with (satellite or aerial) images. One application for such integration could be for the purpose of automatic recognition and annotation of spatial objects in imagery. We utilized a wide variety of geospatial and textual data available on the Internet in order to efficiently and accurately identify objects in the satellite imagery. To demonstrate the utility of our technique, we built an application that utilizes the satellite imagery from the Microsoft TerraService and the Tigerline vector files from US Census Bureau (as well as some online sources) to annotate buildings on the imagery. Our main challenge is that geospatial data (specifically, vector and image data) obtained from various data sources may have different projections, different accuracy levels, and different inconsistencies. The applications that integrate information from various geospatial data sources must be able to overcome these inconsistencies accurately, in real-time and for large regions. Traditionally, this problem has been in the domain of the image processing and GIS systems. However, the conflation approach used in various GIS systems to manually or semi-automatically align two geospatial data sets does not scale up to large regions. We utilize a fully automatic conflation technique to fix the vector data and image alignment.

Based on the above approach, we use common vector datasets as "glue" to further integrate imagery with maps. This is done by having a set of control point pairs for the map and imagery, using conventional conflation technique to align the map with the satellite imagery. Please refer to Geosemble Technologies for more details.

                      

 

Effective Presentation and Querying:

We are currently developing a user interface - Negaah- (Negaah in Persian means “viewing”) for GeoDec that will allow the user to navigate and interactively query the 3D environment in real-time. Negaah allows greater decision-making flexibility by allowing the user to query geospatial data based on a user-defined selection area. Negaah will also allow the user to submit queries based on a different time intervals for temporal data (such as object trajectories or stored video streams). The users can selectively query and display different layers of information. Negaah also supports more sophisticated queries such as calculating the shortest path between two points.


All the queries in Negaah are directed to GeoDec’s information mediator/spatio-temporal database component through a new middleware layer, Jooya (Jooya in Persian means “finder”). Jooya offers a universal way of specifying the type of query, as well as its parameters by a GUI and retrieves the results back in a unified way (currently using the KML format). Jooya queries not only the information mediator but also our private spatial database for vector data, moving objects and 3d building models as well as the video-server. It is a thin layer between GUI and mediator/database/specialized-servers to unify query as well as query results.
Therefore any visualization layer can sit on top of Jooya for its integrated query and access needs. Currently, we could port Negaah, Microsoft Virtual Earth and Google Earth on top of Jooya. This design makes the interface independent of the inherent data model and facilitates scaling the architecture by allowing several visualization components to specify queries and receive the results back in a uniform language hiding the source of information.

Examples of data sources that Negaah can already query include:

  • Temporal 3D buildings (to trace the historical 3D modeling of buildings);
  • Point data (eg. text information concerning a query point, like building names);
  • Point of interests (eg. Restaurants, Named Place, Parkings, Hospitals ...);
  • Road vector data;
  • Traffic information;
  • Moving objects/trajectories (such as trams);
  • Parcel Information;
  • Video;
  • Event (e.g Query about videos concering a given event);
  • Visibility;
  • Nearest neighbors;

 

The following example shows combined queries of "3D buildings" and "Point of Interests":


 Video Query:
In addition to images (maps), alphanumeric and geospatial vector data, we are integrating numerous video streams acquired from video sensors that are accessible within the GeoDec environment. We provide the functionality to search for the videos within a given area of interest and during a given time interval. The video sources can either be spatio-temporaly indexed videos stored in media servers or live videos acquired from live cameras. The video clips satisfying the query conditions are streamed and displayed to user. Our media streaming and storage architecture is specifically designed to scale to a large number of concurrent streams. Users might simultaneously view several live and pre-recorded videos in a geospatial region and go back in time by navigating through a temporal history browser implemented in Negaah. We have enabled this functionality by continuously recording a sliding temporal window of all the incoming video streams. 

 

Visibility Query:

The visibility query color-codes all the areas (including gournd and building facades) that is visible to/from to a query point within a certain range.  

Blue: Building facades and grounds that are visible to the query point.

Glove-Based User Interface
We have developed a glove-based user interface for users to navigate these immersive and information-rich three-dimensional environments.
Our system interprets user commands based on hand gestures obtained using data gloves and an extended range tracking device. This interface allows the user to navigate and interact with the environment intuitively, particularly when using a large display.
In addition, automatic level of detail control displays point data (e.g. building names) on the model when it is contextually relevant to the user, based on the distance of the user from the object. The user can also use the glove-based user interface to view other information integrated with the model such as conflated vector or image data.

 

Challenges:

  • Realistic rendering
  • Accurate information fusion
  • Interactive query and access
  • Scalable infrastructure
  • Efficient in time-to-build

 

Applications:

  • City Planners
  • Military intelligence
  • Simulation & training
  • Computer game
  • Real-estate
  • News broadcast

 

Future work:
As part of our future work we would like to achieve the following goals:
  • Real applications, multidisciplinary collaboration
  • Dynamic datasets: climate data, transportation data
  • Marrying
    • Information richness of GIS systems
    • Interaction flexibility of computer games & simulations

Download a copy of GeoDec Brochure from here.

 

 

 

 

Sponsors: