Research On The Approach To The Adaptive Visualization Of Large-scale Forest Scene

Forest growth simulation needs to consider the dynamic growth features and the environment influences to predict the growth trend of the plants and displays the growth status of the forest to the users. Forest is a complex system with wide distribution and lots of plants, which not only needs to store the information of every plant, but also needs to compute the interactions between the plants. It is time-consuming and resource-consuming to simulate the growth process for large-scale forest scene, thus greatly restricting the application and research of the forest simulation.Although with the fast development of the3D graphics techniques, it cannot meet the increasing demand of the large amount of scene data, and even the top-of-the-range hardware computers cannot display all the scene data in real time. In the real-time visualization of any3D complex scene, how to automatically adapt to the different requirements of the users and software-hardware environments to generate satisfied visual scenes has already become a hotspot of3D interactive game, digital city and virtual forest simulation. There is currently very few researches carried on in the field of adaptive visualization of large scale forest scene. In order to automatically adapt to all kinds of terminals that have different software and hardware configurations, this paper proposes a time-critical adaptive generation approach of interactive3D complex scene, which synthetically consider the computing performance of computer, rendering time and visual effects. Therefore, it is meaningful to emphasize the practical application of forest simulation. The major work and achievements are as follows:(1) In order to solve the problem that the plant growth computing costs lots of time in large-scale forest scene simulation, this paper proposes the approach to the fast generation of large-scale forest scene based on the spatial similarity. It makes use of the property that the plant may have similar growth environment with the plants adjacent to it. The approach accelerates the simulation process of large-scale forest scene and improves the simulation efficiency in resource-constrained computers.(2) In order to allocate the resources with time as the leading factor in the adaptive visualization of forest scene and keep the simulation time in the user-specified time, this paper proposes the approach to estimating the simulation time in advance. With the increase of scene complexity, an increasing proportion of the texture is used in the visual model of the forest scene. To get more accurate estimation time, this paper considers both geometry and texture rendering time.(3) In order to select the models with best visual contribution to the scene visualization, this paper proposes the approach of evaluating the object importance. The evaluation of object importance is a crucial aspect to realize adaptive visualization. When the estimated rendering time exceeds the specified time set by the user, it can shorten rendering time by reducing the LOD levels of the relatively unimportant trees. When the estimated rendering time is less than the specified time, it can improve the LOD levels of important trees to increase rendering time, thus making the rendering time stable. Then object importance computing method in this paper can estimate the object importance intuitionally and scientifically, which is the basis of selecting proper model.(4) On the basis of investigating the development of adaptive visualization of forest scene, it constructs the interactive adaptive visualization system of large-scale forest scene. The framework of the adaptive rendering consists of user interaction layer, system core layer and data layer. User interaction layer provides the interactive functions such as platform parameter initialization, scene model management, scene visualization and performance analysis management to the users. System core layer mainly comprises two modules of plant growth computing and the adaptive visualization. Data layer focuses on the data access logic whose aim is to isolate the data access and data source.(5) This paper constructs the adaptive visualization of large-scale forest scene to verify the correctness of the approaches to adaptive visualization. This paper compares using the time-critical adaptive approach and not using time-critical adaptive approach in the aspects of average frame rate, the number of LOD. Further, we have tested on different configurations of computers. The application shows that the method can finely adapt to computers of different computing abilities and system resources, keep the realism of forest scene, and achieve rather good visual realism and persistent frame rate.