Geo. Dec: Enabling Geospatial Decision Making Faculty Cyrus Shahabi (DB) Craig Knoblock (AI) Ulrich Neumann (CG) Ram Nevatia (CV) Staff Farnoush Banaei-Kashani Luciano Nocera National Science Foundation Engineering Research Center University of Southern California Los Angeles, CA 900890781 shahabi@usc. edu http: //infolab. usc. edu Ph. D Students Ugur Demiryurek Jeff Khoshgozaran Songhua Xing Undergrad Student Fernando Arreola
Geo. Dec Mission! Vision (What? ): • Rapidly and accurately building an information-rich and realistic geospatial space (e. g. , a city) with temporal dimension, supporting visualization, querying and data analysis capabilities Applications (Why? ): • City planners • Challenges (How? ): • Emergency response and first – Realistic rendering responders – Accurate information fusion • Military intelligence – Interactive query and access • Simulation & training – Scalable infrastructure • Computer games – Efficient in time-to-build • Real-estate • News broadcast National Science Foundation Engineering Research Center
Geo. Dec Data Sources National Science Foundation Engineering Research Center
Outline • Underlying Technologies – – Building 3 D Models Static and Dynamic Textures Geospatial Data Fusion Effective Presentation and Querying • Current Status and Future plan National Science Foundation Engineering Research Center
Outline • Underlying Technologies – – Building 3 D Models Static and Dynamic Textures Geospatial Data Fusion Effective Presentation and Querying • Current Status and Future plan National Science Foundation Engineering Research Center
3 D Building Modeling Overview (Nevatia et. al) National Science Foundation Engineering Research Center
Large Site Modeling Example (USC Campus) 3 D models with texture rendered in VRML 3 D building models rendered in VRML • 256 building components modeled in about two hours National Science Foundation Engineering Research Center
Example models National Science Foundation Engineering Research Center
Outline • Underlying Technologies – – Building 3 D Models Static and Dynamic Textures Geospatial Data Fusion Effective Presentation and Querying • Current Status and Future plan National Science Foundation Engineering Research Center
Textures for building facades (Neumann et. al) • Aerial views provide streets, roofs, and open spaces (parks, plazas, etc…) • Ground views capture building facades – Occlusions from landscaping, poles, architectural protrusions (e. g. , entry canopies) need to be removed – Requires fusion of images from multiple perspectives to see behind occlusions – methods to reduce parallax artifacts – Alternatively, portions of the textures need to be synthesized based on building-specific texture patterns App 1 -The Grand Ave. Project: “Allow people to explore the physical location of the Grand Avenue park virtually. ” See: http: //www. learcenter. org/html/about/? &cm=grand/3 d National Science Foundation Engineering Research Center
Social Image Mapping (SIM) • Improving time-to-build • Collaboration with NSF’s STC at UCLA: CENS: Center for Embedded Networked Sensing National Science Foundation Engineering Research Center
Social Image Mapping (SIM) National Science Foundation Engineering Research Center
Social Image Mapping (SIM) National Science Foundation Engineering Research Center
Dynamic Textures from Video Streams App 2 -The DC Project: Five video stream fusion on 3 D models in an area at Washington DC National Science Foundation Engineering Research Center
Outline • Underlying Technologies – – Building 3 D Models Static and Dynamic Textures Geospatial Data Fusion Effective Presentation and Querying • Current Status and Future plan National Science Foundation Engineering Research Center
Goal: Vector, Map and Image Data Fusion (Knoblock & Shahabi) TIGER/Line Vector Data Satellite Imagery Automatic Vector to Imagery Conflation Tram Route/Stop Map Automatic Map to Imagery Conflation Aligned Vector Data overlapped Aligned map overlapped with the 3 D model Model App 3 -The Grand Ave. Project: “ 3 -D model of the proposed park area that has been enhanced with maps, road information, still photos, panoramic images and historical information about the unique art and architecture in the area. ” See: http: //www. learcenter. org/html/about/? &cm=grand/3 d National Science Foundation Engineering Research Center
Demo: Sponsors: • NSF – IMSC – ITR – PECASE • • Google Annenberg Microsoft Chevron App 4 -The USC's 125 th Anniversary: Adding moving objects and glove interface National Science Foundation Engineering Research Center
Outline • Underlying Technologies – – Building 3 D Models Static and Dynamic Textures Geospatial Data Fusion Effective Presentation and Querying • Current Status and future plan National Science Foundation Engineering Research Center
Query Formation/Visualization – Negaah (Shahabi et. al) Negaah – New User Interface • Supports formation of queries with user-defined spatiotemporal parameters • Query types: – Range • Spatial • Temporal • Spatiotemporal – – – shortest path nearest neighbor trajectories events objects (3 D models, parcels, …) – Line of Sight National Science Foundation Engineering Research Center
MS VE • Range Queries • Video • Vector Data • Buildings • Tram Location • NN Queries • Shortest Path Queries Google Earth GUI interface Negaah Unified Query Results KML Files Jooya Unified Query Darya Query Results Prometheus Yima Video Streams Web pages GIS Datasets Semi Automatic Processes Video Streams Gazetteer Data Satellite/Aerial Images i. MVS Static/Dynamic Textures Conflation National Science Foundation Engineering Research Center Vector/Raster Map Temporal Data
Complex apps Paradigm Shift Arc GIS Scale-up Windows Live Local Geo. Dec Arc Explorer Google Earth Analysis & Process at Analysis & Process distributed Analysis & Process at Applications between application & data Servers Data Servers National Science Foundation Engineering Research Center
So What’s unique about Geo. Dec? • Not a visualization GUI with data access features added in an ad-hoc manner as an after thought! – Three tier architecture with distribution of code where it belongs – All objects are tagged and indexed spatially and temporally – Access at different tiers: GUI, Web-services, database • Time dimension • Extended querying capabilities • We have the source code! – Go wild! Your ideas will not be hindered by the restricted API! National Science Foundation Engineering Research Center
Outline • Underlying Technologies – – Building 3 D Models Static and Dynamic Textures Geospatial Data Fusion Effective Presentation and Querying • Current Status and future plan National Science Foundation Engineering Research Center
In the News… Lectures National Science Foundation Engineering Research Center
Where we’re going? • Up to now, the focus was on: – – – Modeling and visualization Acquisition, storage, integration of real datasets Implementing the software architecture Developing generic query types Some “toy” applications: DC, USC campus, downtown LA • What’s next: – Geo. Dec+: Challenges of • Real applications, multidisciplinary collaboration (e. g. , NSF CDI pre-proposal: pollution in Megacities w/ Keck School of Medicine, Env. & Civil Eng, Earth Science; School of Architecture) • Dynamic datasets: climate data, pollutant data, transportation data, … National Science Foundation Engineering Research Center
Geo. Dec+: Vision • Marrying – Information richness of GIS systems – Interaction flexibility of computer games & simulations • Challenge: Query and access of real-world large spatio-temporal data as if they are synthetically generated data – Index 3 d data (e. g. , surface data) for querying and not just rendering National Science Foundation Engineering Research Center
Thanks! National Science Foundation Engineering Research Center
Скачать презентацию Geo Dec Enabling Geospatial Decision Making Faculty Cyrus
38d6087dae6697183359a3857b3ced95.ppt