Research On The Real-time Building Technology Of Mobile Phone Test Dynamic Scene

In recent years, quick modeling technology of large-scale3D scene is widelyused in the field of three-dimensional visualization. It not only provides the technicalsupport for urban planning, intelligent transportation, digital entertainment, militarysimulation and other fields, but also has been widely used in people’s daily life, suchas digital earth, virtual tourism, three-dimensional navigation, etc. so it has veryimportant value on research and application.In order to build virtual test environment of mobile dynamically, this articledevotes to researching the dynamic construction technology of large-scale3D scene.Because of wide modeling range and high updating frequency, the requirements forthe construction efficiency and the automation degree of virtual scene become moreand more higher. In view of the lack of previous modeling method, considering theimage data can express different scenario simply and be easy to compress and store,this paper proposes a kind of procedural modeling method based on image rules. Themethod is on basis of existing2DGIS data, through the establishment of the imagerules, in the form of image saves the geometry character of each element in the scene.And then through the generated rules of height, this method gains the heightinformation of each ground object model. At last, using the modeling method basedon triangular subdivision, this method constructs a large-scale3D scene in accordancewith certain accuracy. This article proposes the corresponding solution for the keyproblems involved in the realization process of this method, such as the separation ofmodel data, automatic generation of the height, the construction of different groundobjects model and so on. In order to realize the dynamic construction of3D scene, thispaper also puts forward a reasonable and effective organization and scheduling policyof the scene data, updating and building the scene in real time, finally achieves thereal-time roaming of large-scale3D scene.This article tests the algorithm with several scenes in Beijing. From the testresults, the modeling method in this paper not only takes up less space, has a high datautilization rate, but also can be realized simply and rapidly and has certain authenticity.So it’s very suitable for dynamic building of large-scale3D scene. Moreover, theimage processing can be extended to parallel computing, which lays the foundation ofhardware implementation for fast modeling. Therefore the algorithm proposed in thispaper has very important theoretical and application value.