Kyong-Ho Kim, Kiwon Lee and Jong-Hun Lee
Image Processing Dept. Electronics and Telecommunications Research
Institute
(formerly Systems Engineering Research Institute)
P.O. Box 1, Yusung, Taejon, 305-600, Korea
Email: {khkim,kilee, jhlee}@seri.re.kr
Recently, the Geographic Information System (GIS) based on 3-dimensional geoprocessing technology and Internet environment are regarded as one of emerging issues in the GIS fields. We attempt to develop a prototype 3D GIS with consideration of the strategic linkage of Java and Virtual Reality Modeling Language (VRML). In this version, several key concepts of state-of-the-art GIS are studied: designing of 3D feature format, handling of spatial features and/or aspatial attributes and multimedia data, spatial indexing of selected feature, developing of 3D analytical or metric operators such as 3D buffering function, near function, and distance or statistical measurement. Especially, concept of "Lantern Operator" in 3D geographical analysis functions is newly devised, and it is basically controlled by floating node, 3D location input of source point, and its direction, effective-length and azimuth.
The integration of virtual environment and GIS has been initiated [1]. Since the mid 1990s, 3D GIS on World Wide Web (WWW) [2] has been regarded as one of promising alternatives in the GIS field mainly due to cost-effectiveness and wide accessibility. In this newly emerging approach, Virtual Reality Modeling Language (VRML) [3] shows several linked aspects with 3D GIS under Internet environment [4]. Actually, there are several attempts to use VRML for cartographic presentations and modeling [5]; however, it was confined to displaying dynamically pre-formatted VRML files by other authoring tools in a certain VRML browser. Accordingly, they are not, in some extents, satisfied with fundamentals or key components of GIS: manipulation of real-coordinate spatial data, GIS-type file conversion or transformation, and spatial analysis and so forth. Functionalities for 3D geographical analysis are those of the most important parts in 3D GIS. 3D geographical analysis can be categorized by their functionality: geometric analysis, spatio-relational analysis, geometry-generating analysis. Geometric analysis is performed on one geographic object, for example, calculating volume and surface area of an object. Spatio-relational analysis is performed on two objects to find the relationship of two objects in 3- dimensional space. There are analyses such as enclose, meets, and nearest. Geometry-generating analysis is performed on one or two objects and produces at least one object as a result of the analysis. In this analysis, are there 3-dimensional buffering, intersection, merge. The design and implementation of the operators for these 3-dimensional geographical analysis are so closely related to the modeling scheme of geographic objects besides the complex computational geometry algorithm itself. The more amount of data is used for modeling an object, the deeper level-of-detail (LOD) can be obtained. But the level of detail should be traded off with the availability in managing the geographic vector data and transmission efficiency in web-based application. We used rather simple modeling scheme for 3-dimensional geographic objects using mainly extrusion method. With this method, 3D objects such as building, road, pipeline can be simply modeled using 2-dimension vector data. The spatial operators implemented in this system is based on this simple modeling strategy but can be utilized for cost-effective 3D analysis. Examples on this operators including 3-dimensional buffering will be shown in the following chapter. For more quantitative spatial analysis, wireframe-like modeling of geographical object is required in addition to the sophisticated computational geometry algorithm. It may be preferable to apply both simple and cost-effective qualitative analysis and complex and costly quantitative analysis method, in accordance with the LOD scheme of the 3D GIS. "Lantern operator" is newly devised in this system. It can be applied for 3-dimensional line-of-sight analysis. This operation is performed on an object and generate lantern-flash-shaped geometry. Some kind of analysis mentioned before, such as the volume, enclose, and intersection can be performed additionally with this operation.
Using External Authoring Interface (EAI) [6], it becomes possible to link external applications to the 3D VRML scene. EAI is a kind of Application Programming Interface (API) to allow the Java applet or application to interact with the VRML scene. This interactivity enables Java applet to build and update dynamically the data in VRML. This nature of EAI makes itself to be applied for various field using dynamic visualization [7]. We take the EAI as the method for our web-based 3D GIS application for three main reasons. First, with EAI, we can utilize the functionality of VRML browser featured mainly by the navigation functions. A toolkit approach such as Java3D [8] requires a programmer to understand the structure of any object imported and to have sophisticated graphic design ability. It can also make it extremely tedious to create and control 3D contents. Second, VRML file can be dynamically built and updated via the EAI, based on data received by Java applets, and in turn, the applet's data can also be dynamically updated through the VRML interface. Using this property, we can have the web-based 3D GIS engine such as 3D GIS data handling module, and 3D spatial operators built by Java applet communicate with the 3D geo-spatial world built by VRML files. Third, the web-based 3D GIS applet utilizes EAI and VRML can easily be accessed by any platform having only Java-enable web browser with VRML browser plug-in. Supporting for the low-cost and platform-independent client is one of the main features of this system.
The overall system diagram for 3D geographical analysis is shown in fig. 1. Data for 3D geometric objects are stored in database. This system is feature-based one, that each object (instance of a feature) has its own methods defining its behavior in addition to geographic data. Retrieval and saving of data is managed by Data manager. It also performs filtering on data using 3D index module. Geographic data are transferred to corresponding spatial operator and on which 3D spatial operation is performed. Managing on each operator is done by Geographical analysis manager. Another role of this module is to communicate with VRML interface called EAI. It sends the result of spatial analysis to EAI having VRML browser visualize the 3D geographic object. It also receives events occurred in VRML browser via EAI, and processes them. The result of spatial analysis may also be presented in textual (numerical) or graphical form using Java Graphic User Interface (GUI).
Figure 1. Overall system architecture.
Some 3D spatial operations are devised and simulated. 3-dimensional buffer operation and near operations are performed on 3D features and visualized. A newly devised operation named "Lantern" operation is simulated on virtual world. Except for the metric operation such as distance measure, the other operations shown in this paper are for GIS-style spatial analysis focused on the visualization.
Each 3-dimensional feature visualized in VRML world has sensor detecting the mouse input. Event handler catch the events from sensor and identify the 3D feature on which the event occurred. Then the feature is consulted for attributes, sounds, and images through each output module. Event handler also alter the appearance of selected feature for user notification. Fig. 2 shows the aspatial attributes and multimedia information (sound and image) related to a school feature selected by user.
Figure 2. Aspatial attributes and multimedia information related to a school feature. Sound can be heard when the "Sound" button is clicked.
Buffer operation for 3D geographic features is available for a lot of 3D spatial query and thus, plays important role in decision making process. 3D buffering on point-shaped feature can be applied for the analysis of 3-dimensional pollution area caused by point sources. Analysis process is visualized using Sphere node of VRML. 3D buffering on line-shaped feature can be used when decide position for laying the water-pipe, gas-pipe, electric cable, or telephone cable under ground. Buffering on 3D polygon-shaped feature can be utilized for city-planning and landscape architecture simulation. The visualization for buffering on 3D line-shaped feature and polygon-shaped feature is implemented via Extrusion node of VRML. 3D buffering on underground water-pipe feature is shown by fig. 3.
Figure 3. 3D buffering on pipeline feature under ground.
Near operation is to find feature(s) located within an extent from a source feature. For example, a query to find all buildings within 100m area from selected point (or feature) is possible (Fig. 4). Near area and features of query result is visualized by coloring. Attributes of those features can also be consulted.
Figure 4. Near operation on a 3D building feature.
Lantern operation is quite newly introduced in this research; using this operation, 3D geographic and geoscience application is possible, beyond limitation of conventionally static 2D GIS. This operation can be regarded as 3-dimensional expansion of 2D line-of-sight analysis, and thus can be applied for 3D viewshed analysis, UHF's or VHF's audible range analysis, and decision for optimal position of transmitting tower. When the source position, effective length, effective angle, and the orientation is specified by user (Fig. 5), Lantern operation can be performed. From the result of Lantern operation, qualitative 3-dimensional visual analysis can be made. For more quantitative analysis, the intersection operation between 3D objects and Lantern-flash can be followed. Applying Lantern operation on a point of pipeline under ground is shown by fig. 6, and the effect on above ground is shown by fig. 7.
Figure 5. Input parameters for Lantern operation.
Figure 6. Lantern operation performed on a point source under ground.
Figure 7. Result of Lantern operation showing the effection to 3D objects above ground.
In this paper, we applied VRML, Java, and EAI technologies to design and implementation of 3D GIS focused on 3D geographical analysis. Although 3D GIS itself is now research-based stage, not reached commercial market stage yet, this approach can demonstrate some advantages over conventional GIS's of 2D GIS, 3D CAD-based data generating system, and even Web mapping system: cost-effectiveness through utilization of VRML browser, dynamic interaction between VRML and Java, and public accessibility of platform independence and 3D spatial operation functionality. As for 3D GIS operator, analytical functions such as 3D buffer and near are implemented. Moreover, "Lantern" operation is newly devised in this research, and it is expected that wide 3D GIS applications using this operator can be realized in future. Whereas, GIS's metric functionality dealt with real geo-coordinate system is included in this system, as normal GIS component. As further works in this approach, algorithms of computational geometry are necessary for displaying of complex 3D geographical objects facing in our real-world, building of 3D data structure, and devising of 3D GIS operators, and it is still progressing.
[1] Neves, N. et al., Cognitive spaces and metaphors: A solution for interacting with spatial data, Computers & Geosciences 23(4), 483-488, 1997.
[2] Rhyne, T. M., Going virtual with geographic information and scientific visualization. Computers & Geosciences 23(4), 489-491, 1997.
[3] VRML, The Virtual Reality Modeling Language (VRML) 2.0 Specification, http://vrml.sgi.com/moving-worlds/spec/part1, 1996.
[4] Fairbairn, D., The use of vrml for cartographic presentation. Computers & Geosciences 23(4), 475-481, 1997.
[5] Tilmann, R., A world of worlds. http://www.meshmart.org/wow/, 1997.
[6] Marrin, C., Mccloskey, B., Sandvik, K., Chin, D., Creating Interactive Java Applications with 3D and VRML, Silicon Graphics, Inc., http://cosmo.sgi.com/products/player/developer/eai, 1997.
[7] Kimen, S., The New Dimension of Visual Finance. Transaction Information Systems, http://vrml.sgi.com/features/ticker/ticker.html, 1996.
[8] Deering, M. F., Sowizral, H. A., Frequently Asked Questions Java3D. Sun Microsystems, Inc., http://java.sun.com/products/java-media/3D/forDevelopers/java3dfaq.html, 1997.