A Cost-effective Strategy for Multi-scale Photo-realistic Building Modeling and Web-based 3-D GIS Applications in Real Estate

Abstract

Web-based 3-D GIS may be the most appropriate tool for decision makers in land management and development. It provides not only the basic GIS functions, but also visually realistic landscape and architectural detail. It also gives the user an immersive 3-D virtual reality environment through the Internet that is rather different from that obtained merely through text, pictures, or videos. However, in terms of high accuracy and level-of-detail (LOD), the generation of a fully photo-realistic city model is labour intensive and time consuming. At the same time, from the aspect of computer graphics, the result is simply a geometric model without thematic information. Thus, the objective of this study is to propose a cost-effective multi-scale building modeling strategy based on the 2-D GIS building footprint that has rich attributes and to realize its application in the real estate market through a web-based 3-D GIS platform. Generally, the data volume needed for a photo-realistic city model is huge, thus for the purpose of increasing Internet data streaming efficiency and reducing the building modeling cost, a multiple-scale building modeling strategy, including block modeling, generic texture modeling, photo-realistic economic modeling, and photo-realistic detailed modeling is proposed. Since 2-D building boundary polygons are popularly used and well attributed, e.g., as to number of stories, address, type, material, etc., we are able to construct the photo-realistic city model based on this. Meanwhile, the conventional 2-D spatial analysis can be maintained and extended to 3-D GIS in the proposed scheme. For real estate applications, a location query system for selecting the optimum living environment is established. Some geospatial query and analysis functionalities are realized, such as address and road-junction positioning, terrain profile analysis, etc. An experimental study area of 11 km² in size is used to demonstrate that the proposed multi-scale building modeling strategy and its integration into a web-based 3-D GIS platform is both efficient and cost-effective.

Key words: Web-based 3-D GIS, Multi-scale Building Modeling, Land Management.

l Results of multi-scale building modeling

Examples of the photo-realistic detailed model, photo-realistic economic model, generic texture model, block model, and bird-eye view images are shown in Figures 4 to 8. In Figure 4, one can see that the photo-realistic detailed model can produce a comparably detailed model in geometry when compared with commercial 3-D modeling software. In Figure 5, the high-resolution texture of the photo-realistic economic model allows for detailed identification of characters and signage, which demonstrate the potential for further applications. The time consumption and man-power needed to perform the photo-realistic detailed model is about 3~4 times the economic one. The amount of nodes and textures between them is also the same ratio. Figure 6 shows the buildings that are not along the main roads, described by the generic texture model. The purpose of this model was not to produce high quality photo-realistic facade textures but to provide images of sufficient realism to provide comfortable visual effect for the user. Figure 7 shows the created block model, which was located outside of the study site. Figure 8 shows a bird-eye view image demonstrating the produced multi-scale building models for the whole study area. It demonstrates that the model developed in this study is able to provide users with a photo-realistic visual experience.

It took about 360 man-days to reconstruct the whole city model. The total number of buildings in Anping is 14,500, but only 3,500, i.e., about 1/4 of the total, were reconstructed as photo-realistic economic and detailed models. The remainder, such as the generic texture models can be reconstructed fully automatically. This means that with the proposed multi-scale building modeling strategy, we can save 3/4 of effort to complete modeling of the whole city with optimal visual effect and all possible GIS functionalities. Meanwhile, from the point of view of efficiency, with the system the client can perform an interactive fly-through with 9-27 frames/sec. Even in a wireless Internet environment, the 3-D scene rendering and spatial query usually can be completed within 27 seconds. Figure 3 indicates the distribution of various scales of building models used in the study.

Figure 3. Distribution of various scales of building models

l GIS query and spatial analysis

Two types of GIS techniques are implemented in this study, namely the query and spatial analysis. In general queries, we establish both text-based geographical positioning and graphics-based geographical entity attribute queries. Thus, one can query a specific location by means of the address, road name, road junction, landmark, etc., with geographic coordinates stored in the database. One may also click anywhere during 3-D browsing. One can then proceed with queries such as the Living function facility, NIMBY, Special facility, and Other facility. An example of querying the Living function facility is illustrated in Figure 9. In this example, four Living function facilities are found within a radius of 250 meters from the target building, including a park, a post office, a kindergarten, and a high school. It seems that the target building would be a high quality living environment for raising a child.

Four physical environment analysis functions of spatial analysis (i.e., shadow simulation, view shed simulation, terrain variation analysis, and sea level rise simulation) were designed to assist the user in the evaluation of a suitable living environment. In the meantime, several measurement tools and a route planning function were developed to assist the user to estimate areas of interest. Figure 10 demonstrates a simulation of a rise in sea level of three meters. In this scenario some buildings and roads might be flooded (blue region). Meanwhile, one may also analyze variation in terrain elevation to evaluate possible flood-prone areas. Figure 11 shows that the terrain elevation is 50-100 cm lower in the middle of the road compared to points A and A’. This means that water may accumulate in this region after heavy rainfall. If the investigator plans to buy a parking lot for the construction of a building, he/she can utilize the measurement tools to estimate its area (as shown in Figure 12), width and distance to any Living function facilities, etc.