The Integration of ArcView/3D Analyst and 3D Simulation Technologies for Interactive Visualization of Urban Environments

Brian E. O’Toole, ITspatial LLC

Mary Beth Fletcher, Arlington County GIS Mapping Center

Russell Banks, Arlington County GIS Mapping Center

Abstract

The Arlington County Department of Public Works GIS mapping center has embarked on a project to develop a real-time 3D visualization capability that is derived from, and totally integrated with the County’s GIS infrastructure. The objective of the project is to develop the tools and processes that will provide the County with the capability to interactively visualize, in both two and three dimensions, selected areas of interest within the County. With this technology, Arlington County’s GIS Mapping Center will be able to support the needs of various County agencies involved in economic development, community development, and public safety. The following paper will discuss several key elements of this project, which include the production of real-time urban databases, and the development of a real-time 3D viewer with the ability to interactively query and display associated GIS information in ArcView. Specifically, the paper will present the process through which a photographically realistic 3D database of Arlington County was produced utilizing Esri’s ArcView GIS and MultiGen-Paradigm’s 3D modeling tool CreatorÔ. Issues relative to interactive real-time visualization, database query mechanisms, and 2D/3D correlation will also be addressed. Finally, examples of the completed database will be presented.

Introduction

Arlington County’s Department of Public Works GIS Mapping Center and ITspatial have developed an extension to the County’s Esri based GIS to support interactive 2D and 3D visualization of existing and planned urban developments. A 3D photo-realistic database was produced encompassing a fifteen-block area of Rosslyn, Virginia, spanning from the Key Bridge to the Iwo Jima Memorial. Employing state of the art real-time 3D modeling and simulation technologies, the database was constructed from the County’s GIS data and available imagery. Further, a real-time 3D viewer was developed as an extension to ArcView with the ability to interactively query and display GIS information in both two and three dimensions. With this technology, Arlington County’s GIS Mapping Center will be able to support the needs of various County agencies that are involved in economic development, community development, and public safety.

The primary technical objective of the project was to demonstrate the feasibility of integrating high fidelity real-time 3D modeling and visualization technologies within the existing GIS infrastructure of the County. Issues related to GIS database attribution, accuracy, and correlation for 3D visualization were of primary interest. The project was a joint development between ITspatial and Arlington County providing for the development of a solution, which encompassed the following:

§ The production of a real-time photo real 3D database from the County’s GIS source data

§ The development of a real-time 2D/3D visualization solution capability of running on a PC

§ The development of an interactive 3D viewer and query interface as an extension to ArcView

§ The development of a methodology for connectivity between the 3D model and 2D GIS data

The following paper provides an overview of the project addressing two major areas of interest, the production of real-time 3D Urban databases from GIS data, and the development of an extension to ArcView for interactively querying and navigating in both two and three dimensions.

Overview of the Rosslyn Database

A photo realistic real-time 3D database of Rosslyn was developed from the GIS data of Arlington County. The database covers a geographic area of roughly one square mile extending from Key Bridge to the Iwo Jima Memorial. The database includes an accurate representation of the terrain contour with detailed representations of streets, pathways, buildings, vegetation, and culture. As illustrated in Figure 1, imagery and photo textures were used to enhance the 3D model to provide a realistic representation of the urban environment. The utilization of data provided by the County such as maps, GIS, and imagery ensured accurate placement of buildings, streets, pathways, light posts, benches and other cultural features. A site survey was conducted at the outset of the project to fully document the area of interest and identify important features to be included in the database. At that time, additional data, video, and photographs were collected to augment the existing source data.

Figure 1 - A sample image of the Rosslyn Database

Overview of the Database Production Process

The production of the real-time 3D database of Rosslyn was accomplished by employing a proven methodology that has been developed and refined over the years in the course of producing simulation databases for the military and aerospace industries. At the heart of the problem is the need to produce a database that can be efficiently rendered at frame rates conducive for a positive interactive experience. This means achieving a frame rate performance of at least 12Hz on a PC level machine. With the detail and complexity that exists in urban environments this requires that the database be designed and optimized from the beginning with real-time performance in mind. This translates to a problem of scene management where polygon density, database hierarchy, level of detail management, and texture management are critical to meeting real time objectives. For the most part, the ability to build and render the database for real-time performance is what differentiates modeling and simulation tools from companies such as MultiGen-Paradigm from other products that have typically been developed as extensions to the GIS. The best solution at this point is to combine the inherent elements of the GIS with the tools and processes employed within the real-time simulation community.

Although previous applications of this technology revolve around flight simulators and large-scale terrain databases the process is essentially the same. The primary difference for urban GIS applications is in the resolution and density of the source data surrounding the area of interest. For the most part, the data that exists in most GIS systems has very good resolution and is significantly more accurate than the data used in military applications. The production process is very much the same but the potential for very accurate and detailed databases is much higher and can be achieved with the introduction of an efficient methodology for source data management and 3D modeling. For the Arlington County project the processes and steps that were employed included the following:

§ Source Data Collection, Correlation, and Correction

§ Base Terrain Skin Generation

§ Building Placement and Extrusion

§ Detailed Modeling and Texturing

In the following sections the issues related to each of these processes will be discussed and examples of the intermediate data and database products will be presented.

Source Data Collection, Correlation, and Correction

The production and maintenance of high fidelity geo-specific urban databases has traditionally been an expensive and time-consuming process. Most of the cost in producing these databases can be attributed to the lack of data of sufficient quality. The uncorrelated nature of available data requires users to correlate elevation data, vector representations of feature data, and available imagery before the generation of the database can begin. This correlation process can be labor intensive and is typically not scaleable. Fortunately, most GIS organizations have a wealth of data including contour maps, LIDAR, road networks, building footprints, and orthographic imagery. In one form or another all of these sources can be used to build an accurate 3D database. The primary issue is that most GIS databases were not designed with 3D-database production in mind. Therefore some level of correction and additional attribution is necessary for the development of an accurate 3D model. Ultimately, if 3D becomes a common element of the GIS then a strategy for incorporating 3D information into the database design standards should be considered. In the case of the Rosslyn database, the following source data, illustrated in figure 1, was available and employed during the production process:

§ 5ft contour (elevation) data derived from orthographic photos

§ LIDAR data – 1ft elevation resolution at 5ft postings

§ Building footprints stored as layers in the GIS – originally digitized from Mylar

§ Shorelines stored as a layer in the GIS

§ Sidewalk and road networks stored as a layer in the GIS

§ Digital photos of buildings, parks, and cultural features

Figure 2 – GIS Source Data Utilized in the Production of the Rosslyn Database

In general, this data was sufficient for supporting the production process. Minor corrections and modifications were recommended and implemented prior to commencing the base terrain modeling phase. As an example, some of the source data corrections included the squaring of building footprints, the attribution of buildings with height information (extracted from LIDAR), and the interpolation of elevation points between contour maps and the LIDAR data. By refining the data early on significant levels of correlation and accuracy were achieved and maintained throughout the production process.

Base Terrain Skin Generation

The generation of the base terrain skin was the first step in the actual database production process. The objective was to build an integrated polygonal mesh, which accurately represents terrain slope and elevation, location of road networks, and location of shorelines. Base terrain skin is typically generated as a triangular irregular network (TIN). This approach provides effective spatial organization of the data and structurally supports level of detail management schemes for consistent polygon processing. In the production of the Rosslyn database, a constrained triangulation algorithm was used to tessellate the terrain. The algorithm takes as input elevation data, street edges and shoreline edges from the GIS. By defining control parameters, algorithm constraints can be placed on the generation of the mesh. Further, by constraining the mesh to the street edges in the GIS data, roadways are efficiently built into the terrain skin. The resultant mesh, illustrated in Figure 2, is stored in the industry standard OpenFlightÔ format. This open standard, developed by MultiGen-Paradigm, Inc., provides the basis for the real-time structure and organization of the total 3D database.

Figure 3 – Base Terrain Mesh of Rosslyn

Building Placement and Extrusion

With an accurate terrain skin in place, the integration of building footprints and the insertion of building geometry was a relatively straightforward process. Since time was spent early in the process correlating the data in the GIS, building footprints fell into place very nicely and were geo-spatially accurate. For the generation of building geometry several modeling options were available. For this project, 3D analyst was used for the tessellation of basic building geometry. A VRML import utility was used to ingest the building models into CreatorÔ, MultiGen-Paradigm’s popular real-time 3D modeling tool. Once the model was in CreatorÔ, extensive geometric details were added and photo-real textures were applied. At this point in the process, the inherent limitations of 3d information stored in the GIS had been reached. As an example, details surrounding the underlying building infrastructure including plazas, common areas, and garages were not maintained in the GIS. The usable information within the GIS was exhausted and the 3D data associated with the details of the environment were derived or extracted from photographs, CAD drawings, or blue prints. This aspect of the production represents the transition into the detailed modeling and content creation phase.

Detailed Modeling and Texturing

The final step in the production process is the detailed hand modeling of culture and the application of photo textures to the building and site models. This phase of the project represents the creation of content, which ultimately provides the compelling realism of the database. Since this aspect of the production involves a creative element, it is important to understand the objective of the database, essentially this phase can continue indefinitely. For the Rosslyn database, Creator was used to model and texture the following database components:

• Roads, curbs, and sidewalk surfaces
• Trees, light poles, streetlights, and benches
• Detailed Geometry on buildings
• Overpasses and park areas

As part of this process a library of OpenFlight models were developed. With this library in place, common models of features such as lampposts, signal lights, trees, and parking meters could be replicated and utilized throughout the database.

ArcView Plug-in for Interactive 2D and 3D Viewing

An extension to ArcView was developed providing a seamless extension of the GIS to support fully interactive 3D visualization. The goal was to extend the GIS interface with a 3D view that is correlated with the 2D information presented in ArcView. With this extension, the user is able is navigate in real-time the photo realistic 3D database and interactively query information in both two and three dimensions. The key attributes of the extension are:

• The 3D view functions in parallel with the standalone GIS tool (ArcView)
• The selection of features is fully correlated between all three data views – tabular, 2D, and 3D.

This extension provides the user with the ability to select features in any data view, and will have the selected features presented immediately in the other views.

Major User Interface Functions

The ArcView extension provides the four major functions that are listed below. A brief description of each functional capability is provided.

§ Select Features in ArcView

§ Select Features in 3D

§ Navigate in 3D

§ Configure 3D view

Figure 4 Integrated 2D/3D User Interface

Select Features in ArcView – The normal mechanisms for selecting features within ArcView are maintained with the integration of the 3D visualization extension. Within ArcView, the user may cause one or more features in one or more themes in the ArcView Window to become selected. When a selection is made in the GIS data, this selection is reflected in the 3D visualization with highlight coloring of the object.

Select Features in 3D - At any time, the user can use the mouse to select features in the 3D view, with a simple mouse button click. The visualization application determines the ArcView theme and a feature index number for the indicated feature from information attached to the scene graph, and communicates this theme and feature reference to ArcView. This selection can become the current selection in ArcView, or is available for use in a user script.

Navigate in 3D - Control mechanisms and a motion model were developed to allow the user to easily navigate the eye point within the 3D scene. Through either mouse or keyboard controls the user can move forward, backward, left, right, up and down. Additional controls are provided to control the orientation or rotation of the eye point. A sophisticated motion model allows the navigation to occur at a rate proportional to mouse velocity, providing the ability to tightly control the viewpoint and perspective from any location in the database. Optional collision detection processing is provided to prevent users from flying out of the database or through objects in the scene.

Configure 3D View – A control panel is provided to allow the user to modify the nature and behavior of the visualization window. Parameters such as field of view, time of day, and motion sensitivity can be controlled through the consolidated user interface.

Software Architecture

The 3D viewing extension has been developed to extend Arlington County’s ArcView GIS 3.2. The extension mechanisms available in ArcView are the Esri Avenue scripting language with inter-process communication through DDE. However, recent Esri products are extensible through a COM based architecture, and this has been considered in the design. The initial version of the software provides a user interface panel and a visualization window to the desktop of a single monitor PC computer running the ArcView software. The software architecture is an object-oriented design consisting of three main components including ArcView GIS, a consolidated user interface, and the visualization application. For this project the 3D rendering was implemented using MultiGen-Paradigm’s VegaÔ.

Summary

A photo realistic real-time 3D database of Rosslyn was developed from the GIS data of Arlington County. The database covers a geographic area of roughly one square mile extending from Key Bridge to the Iwo Jima Memorial. The database includes an accurate representation of the terrain contour with detailed representations of streets, pathways, buildings, vegetation, and culture. Imagery and photo textures were used to enhance the 3D model to provide a realistic representation of the urban environment. The utilization of data provided by the County such as maps, GIS, and imagery ensured accurate placement of buildings, streets, pathways, light posts, benches and other cultural features.

The ability to integrate 2D and 3D information has been demonstrated with the development of a 3D visualization extension to ArcView. Further, a methodology for dynamic connectivity between the 2D GIS and 3D database was developed.

The objective of the project were met with the tools and processes in place to provide the County with the capability to interactively visualize, in both two and three dimensions, selected areas of interest within the County. With this technology, Arlington County’s GIS Mapping Center will be able to support the needs of various County agencies involved in economic development, community development, and pubic safety.

Acknowledgements

The authors would like to acknowledge Will Richards and Dave Unrau of ITspatial for their contributions to this paper and the outstanding effort that went into the development of the database and the ArcView extension.

Author Information

Brian E. O’Toole

Senior Partner, ITspatial LLC

Brian@ITspatial.com

(703) 506-3969

Mary Beth Fletcher

Manager, Arlington County GIS Mapping Center

Mfltec@co.arlington.va.us

(703) 228-3649

Russel Banks

Cartographer, Arlington County GIS Mapping Center

Rbanks@co.arlington.va.us

(703) 228-7568