Research On The Model Of Animation Of Trees Swaying In Wind Field Based On Video-Input Driven

The three-dimensional simulation of natural scenes is an important area of research in computer graphics. It has a very wide range of applications in3D games, film special effects, computer animation, virtual reality and augmented reality. Tree are the most common plants in the natural scene, simulate realistic trees swaying effect will greatly improve the realism of the scene. Due to the complexity of the tree structure and the power system, the way to simulate the swaying effect by physical methods is not a good way. Therefore it is really necessary to propose a new approach to achieve the effect of the trees swaying in real-time.This paper presents a video-based data-driven solutions for the simulation of three-dimensional trees swaying effect. Image processing module is aiming at the video image information extraction and reconstruction of the three-dimensional stationary trees model, using the mixed edge detection and color-based K-means clustering algorithm to tree image segmentation, split trees into trunk and the cloud point of leaves. Then get the wind speed by mechanical analysis and achieve static trees model by the three-dimensional reconstruction. Statistical model module is focused on the construction of statistical model, inference and parameter learning, then use the model to classify and tracking the video sequence. Motion synthesis module is based on the trajectory statistics module to synthesis the movement of trees, branches corresponding displacement loading movements synthesis to the stationary three-dimensional tree models, load leaves information to trees by the cloud point of collection.Using MATLAB platform to realize the sub-module of system and analysis the running time of each module. From which we can tell that the real-time calculation part only account for12%of the running time of the whole system. Meanwhile the introduction of IBR(Image-based Reconstruction) making a more realistic three-dimensional trees. The wind speed sequence which is based on a statistical model fit the natural conditions of the wind field model well, which make the simulation become more natural.