基于径向基函数的点模型自适应插值算法研究

发布时间：2018-09-03 15:28

【摘要】：随着计算机技术的发展,虚拟现实技术逐步渗入人类生活的各个领域。虚拟现实仿真系统是虚拟现实技术的重要应用之一,此仿真系统通过使用虚拟现实技术向用户提供了一种具有高逼真性及高沉浸感的虚拟世界,且可通过必要的设备与此虚拟世界进行交互,其中,物体发生形变是交互操作中的重要表现之一。在对物体的形变仿真中,使用的物体模型主要有网格模型和点云模型。点云模型具有储存小、简单灵活等特点,因此其常被用于物体发生大形变的仿真计算中。然而在使用点云模型进行大形变仿真时,物体模型可能会出现局部稀疏甚至空洞的现象,进而导致物体模型的后期渲染效果不佳,不能满足虚拟现实系统的高逼真性要求。为了解决这个问题,本文提出一种基于径向基函数的点模型自适应插值方法,主要研究工作如下:(1)提出一种基于径向基函数的点模型自适应插值方法。首先为三维点云空间建立基于二次栅格划分的空间拓扑关系及K邻域搜索,检测点云模型中的低密度点,并通过邻域扩展原则提取点模型的稀疏区域;然后在稀疏区域内,根据低密度点及其邻域点构建其矩形微切平面,并在此平面内进行适应性均匀采样;最后根据稀疏区域内的点数据构建径向基函数插值曲面,并通过梯度下降法将新增的采样点向该曲面进行修正,保证模型的连续性与光滑性。(2)将基于径向基函数的点模型自适应插值方法应用于静态稀疏模型以及模型形变的动态仿真过程中。在算法应用过程中,首先将插值方法应用于静态的稀疏模型,通过对静态稀疏模型的插值实验结果表明,此算法可有效地为模型的稀疏区域进行插值,很好地丰富了点模型信息,且该稀疏区域在插值后可与模型的其他区域的密度保持一致,证明了基于径向基函数的点模型自适应插值方法的可行性及有效性;然后研究并设计了一种虚拟模型形变仿真系统,并将基于径向基函数的点模型自适应插值方法集成到此系统中,实现了模型形变过程中的动态插值,很好地解决了点模型发生大形变后局部出现稀疏的问题,保证了模型在形变动态仿真过程中的渲染效果逼真性。
[Abstract]:With the development of computer technology, virtual reality technology has gradually penetrated into every field of human life. Virtual reality simulation system is one of the important applications of virtual reality technology. This simulation system provides a virtual world with high fidelity and high immersion to users by using virtual reality technology. And it can interact with the virtual world through necessary devices, in which object deformation is one of the important manifestations of interactive operation. In the deformation simulation of objects, the main object models used are grid model and point cloud model. The point cloud model has the characteristics of small storage, simple and flexible, so it is often used in the simulation of large deformation of objects. However, when the point cloud model is used for large deformation simulation, the object model may appear the phenomenon of local sparsity or even cavitation, which leads to the poor rendering effect of the object model, which can not meet the requirement of high fidelity of the virtual reality system. In order to solve this problem, a point model adaptive interpolation method based on radial basis function is proposed. The main research work is as follows: (1) A point model adaptive interpolation method based on radial basis function is proposed. Firstly, the spatial topological relation and K neighborhood search based on quadratic grid partition are established for 3D point cloud space to detect the low density points in the point cloud model, and then the sparse region of the point model is extracted by neighborhood extension principle, and then in the sparse region, The rectangular microtangent plane is constructed according to the low density point and its adjacent points, and the adaptive uniform sampling is carried out in the plane. Finally, the radial basis function interpolation surface is constructed according to the point data in the sparse region. The new sampling points are modified to the surface by gradient descent method. The continuity and smoothness of the model are guaranteed. (2) the point model adaptive interpolation method based on radial basis function is applied to the static sparse model and the dynamic simulation of model deformation. In the application of the algorithm, the interpolation method is first applied to the static sparse model. The experimental results of the static sparse model show that the algorithm can effectively interpolate the sparse region of the model. The point model information is well enriched and the density of the sparse region can be consistent with that of other regions of the model after interpolation, which proves the feasibility and effectiveness of the point model adaptive interpolation method based on radial basis function. Then a virtual model deformation simulation system is studied and designed, and the point model adaptive interpolation method based on radial basis function is integrated into the system to realize the dynamic interpolation in the process of model deformation. The problem of local sparseness after the large deformation of the point model is well solved and the rendering effect of the model in the process of dynamic deformation simulation is guaranteed to be realistic.
【学位授予单位】：南昌大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.9;O241.3

【相似文献】