İstanbul Teknik Üniversitesi / Fen Bilimleri Enstitüsü / Geomatik Mühendisliği Anabilim Dalı
3B kent modellemede genelleştirme problemleri ve ayrıntı düzeyi (LoD) kavramı
Generalization problems of 3D city modelling and level of detail (LoD) concept
Yeryüzünün küçültülmüş modelinin iki ya da üç boyutlu olarak ölçekli gösterimi haritalar aracılığıyla sağlanmaktadır. Haritaların anlaşılması ve üzerinden bilgi elde edilmesi için doğru ve belirli tasarım kurallara göre üretilmesi gerekmektedir, bu nedenle tasarım, harita üretim sürecinin önemini bir bileşeni olarak karşımıza çıkmaktadır. Tasarım aşamasında genelleştirme işlemi; haritanın kullanım amacı, konu ve ölçeğine göre en önemli yanları ile yansıtabilmek için gerekli olmayan bilgilerin gösteriminin göz ardı edilirken, bazı bilgilerin vurgulanarak ön plana çıkarılmasıdır. Dijital haritalar ile birlikte yeryüzünün modellenmesi kolaylaşmıştır ve 3B kent modelleme kavramı ortaya çıkmıştır. 3B kent modelleri; kentsel alanların geometrik bir temsilinden öte mekansal detayların hem geometrik hem de semantik özelliklerini aynı ortamda birleştiren ve bunlar üzerinde mekansal analiz yapabilen modeller olarak tanımlanabilir. 3B kent modelleri için oldukça önemli bir kavram olan ayrıntı düzeyi (Level of Detail- LoD); gerçek dünyadaki objelerin özetlenme derecesi veya öncelikli gösterimi olarak tanımlanabilir. Farklı ayrıntı düzeylerindeki 3B kent modellerinin oluşturulması için genelleştrime işlemlerine gerek duyulur. OGC CityGML standardı, farklı ayrıntı düzeylerinde semantik geometrik modelemeyi desteklemektedir. Ayrıca semantik, coğrafi referanslandırma ve web kullanımı desteğiyle CityGML 3B mekansal veri altyapısı çalışmalarında genel bir standart olarak tercih edilmektedir. Bu çalışmada ilk olarak kartografların önemli çalışma ve araştırma alanlarından biri olan genelleştirme konusu, 3B kent modelleme kapsamında incelenecektir. Ayrıca 3B modellerde LoD kavramı, CityGML standartları ve 3B modellemede karşılaşılan problem çözümleri için yapılan örnek çalışmalar incelenmiştir. Tez de ayrıntı düzeyi kavramı ve ÇGVT gösterim seviyeleri ile birleştirilerek semantik genelleştirmenin ileri araştırma konusu olacağı vurgulanmıştır.
Earth's minimized model representation is provided through maps as two or three dimensional. To understand maps and extract information of them, maps must be accurate and made according to certain rules, therefore, design stage is significant. Generalization process is to fall within design stage. The purpose of generalization is to reduce the complexity of a data set with respect to its size and semantic content while retaining the pieces of information, and this information is relevant to the aim of map and for map use. 2D systems are widely used in GIS, but are limited in solving complex problems. 3D city models make it easier for people to understand the spatial properties of urban objects. 2D systems are widely used in geographic information system (GIS), but are limited in solving complex problems. 3D city models make it easier for people to understand the spatial properties of urban objects. The current popular GIS software handles, manipulates, and analyses geographic data in 2D or 2.5D, thus using this system to manipulate 3D data information about real world objects may not be appropriate. Therefore, the 2D GIS (or 2.5D GIS) needs to be extended, i.e. to 3D GIS. Similar to the 2D cartographic maps, the 3D city models be used to integrate various data from different sources for public accessible visualization and many other applications. 3D city models are the basis of many applications and are the platform for integrating city information from different resources. In 3D city modelling, the concept of level-of-details indicates the methods of collecting data for a certain application. The levels of details are of great use to data analysis and mental visualization. Visualization is a complex and important issue in 3D city model applications. Many applications such as urban planning, facility management and personal navigation require semantic information about the city objects besides the geometry models. It is essential to develop methods for 3D city model semantic information modelling, representation, discovering, management, querying and analysis. Several standards have been introduced to enable the 3D model to be used effectively together with its widespread use and to be understood by all. CityGML is one of them, also most used one. CityGML is a common information model for the representation of 3D urban objects. CityGML covers broad thematic fields of city objects, from geometrical band topological to semantic aspects CityGML, the OGC standard to represent 3D city models, can be used to integrate both geometric and semantic information of the city models. CityGML is object-oriented model that support 3D geometry and thematic semantics, attributes and relationships, and offers advanced options for realistic visualization. Nowadays, a rapidly increasing number of companies create virtual 3D city models for use in various markets, such as urban planning, telecommunications, disaster/crisis management, 3D cadastre, tourism, navigation, facilities management, environmental simulations and smart cities applications. Associated with development of GIS related technologies such as Global Positioning System (GPS), and remote sensing, the volume of raw data especially for 3D city models is increased dramatically. The availability of 3D models is rapidly growing. 3D city models are not only geometrically representation of urban areas but also compound geometric and semantic features of spatial details and also 3D city models can be used for spatial analyses. In 3D models there are often constraints that are extremely difficult to ensure on a purely geometrical basis. In order to use 3D city models beyond the task, a generalization is necessary. There are two main reasons can be identified for the necessity of generalization. Firstly the size of the data sets is too large to be processed in the application. Secondly, removing or rearranging those aspects that cannot be handled by the application. Multi-representation databases (MRDB) allow representing the same real world entities in different cartographic databases with their own level of detail. In an MRDB, different views on the same physical objects or phenomena can be stored and linked. Several standards have been introduced to enable the 3D model to be used effectively together with its widespread use and to be understood by all and City Geography Markup Language (CityGML) is one of them, also most used one. CityGML is an open data model and XML-based format for the storage and exchange of virtual 3D city models. CityGML was developed in order to reach a common definition and understanding of the basic entities, attributes, and relations within a 3D city model. City GML illustrates four different aspects of 3D city models which include: semantic aspect, geometry, topology, and characteristics of the appearance. CityGML are the best known semantic standards that are used by these 3D spatial models. Level of Details (LoD) which is an important concept for 3D modelling, can be defined as outlined degree or prior representation of real-world objects. Different LODs are proposed to decrease the complexity of 3D models. Automatic generalization methods are used to create 3D city model in lower LODs from higher LODs and visualize a large region of urban with a reasonable amount of data, which is very important for online 3D city models. Several algorithms have been proposed for generalization of 3D city models for buildings, roads etc. The purpose of generalization is therefore to reduce the size and semantic complexity of a model to a level that can be handled by the application without losing relevant information. Different generalization strategies lead to different outcomes. It is critical to evaluate the quality of the generalized models. The choice of degree of generalization depends on aim and scale of the map. After the generalization processes on different levels, multiple representation data structures are required to store the generalized models for dynamic visualization. Generalization process is to fall within design stage. The purpose of generalization is to reduce the complexity of a data set with respect to its size and semantic content while retaining the pieces of information, and this information is relevant to the task and for users. It is important to find methods to generate lower level of details (LODs) from higher LODs automatically by generalizing city objects on city parcel, building level and facade level, because generalization not only reduces the data volume but also hides the unnecessary details in different scales according to the aim of map. These generalized representations in different LODs should be unified to support dynamic visualization. This research focuses on visualisation of multi-scale 3D city models derived from generalisation methods. In this study, combining the levels of detail in the CityGML structure (LoD) and MRDB representation levels can be useful for creating 3D models. In this study firstly generalization which is major stııdy and research concerned of cartography, 3D modelling will be examined. The thesis aim is describes some requirements of 3D model generalization, then presents problems and approaches that have been developed in recent years. Also LoD concept in 3D modelling, CityGML standards and problems and solutions these are involved in 3D models will be discussed. In conclude the research will suggest a solution for semantic generalization of 3D city modelling.