A methodology for modelling of 3D spatial constraints

Abstract

In this work we demonstrate a new methodology to conceptualise and implement geo-constraints in 3D, which has not been widely explored yet. This is done in four stages from natural language to implementation, in which geometric primitives and Object Constraint Language (OCL) play a crucial role to formulate the constraints. A database including various 3D topographic objects (e.g. buildings, trees, roads, grass, water-bodies and terrains) from CityGML (no constraints yet) is used as a case study to apply the developed methodology. In this research, a first attempt to formulate 3D geo-constraints in OCL is made. Unified Modelling Language (UML) class diagram has been extended with graphical symbols for indicating constraints between classes (in addition to the additional compartment within a class for a class constraint). These constraint expressions can be tested and translated to other models/implementations when the OCL standard is extended with spatial types and operations. During this research, new types of constraints are defined as follows: general-level constraints (applicable to all object sub-classes), parameterised constraints (containing numeric values, e.g. maximum distance), constraints allowing exceptional instances (to resolve cases that have not been defined) and constraints relating to multi-scale representations (to check the consistency between two levels of detail which model the same object). Additionally common sense rules to detect conflicting constraints are specified as well.