当前位置:主页 > 科技论文 > 软件论文 >

树木三维动态数据获取与动画生成技术研究

发布时间:2018-06-29 12:57

  本文选题: + 动态参数 ; 参考:《西北农林科技大学》2017年硕士论文


【摘要】:真实感树动画对于增加游戏感染力与沉浸感有重要意义。近几年,为提高树动画模拟的真实感,基于数据驱动的树动画模拟方法成为研究热点,而当前基于数据驱动的方法存在数据获取设备昂贵、过程复杂,树模型与真实模型差距较大、模拟树木运动形式有限等问题。针对上述问题,本文采用廉价的Kinect体感设备,研究基于运动视频和深度数据的树木动态参数提取方法和基于点云的树三维模型构建方法,然后将树动态参数与真实感树模型输入到树动力学模型,基于模态分析构建动画模拟系统实现数据驱动的真实感树运动模拟。论文主要研究内容和结论如下:(1)利用Kinect V2搭建数据实时采集的试验平台。通过PC工作站控制架设于三角支架上的Kinect设备,同步采集具有标记点的树枝干运动视频流和深度信息流,并采取先写内存、后写磁盘的存储策略保存采集的运动数据,以满足高帧率连续运动数据的需求。试验表明,在阳光条件较弱的室外,试验平台获取的运动视频流和对应的深度信息流的帧率可达30fps(frames per second),采集的运动数据满足后续研究的需求。(2)针对树木真实动态参数获取设备昂贵、过程复杂的问题,提出基于局部相对偏转角度量准则的树枝干振动频率与阻尼比获取方法。该方法通过跟踪树枝干颜色特征标记,提取各树枝标记点运动的二维轨迹;依据跟踪结果构建二维树骨架运动模型,并计算子树枝相对于父树枝的局部偏转角,通过快速傅里叶变换将时域振动信号变换到频域,频域信号的一阶振动波峰对应的频率即为拟获取的树枝振动频率;利用阻尼比与一阶振动频率振幅衰减的关系,采用半波法计算出阻尼比。结果表明,供试树木的振动频率分别为2.4Hz和2.0Hz,树的振动频率为树的固有特征,不随外力的改变而改变;而树在运动过程中受到的阻尼与外力大小、树的分支结构有明显的关系。(3)为解决数据驱动树动画的真实树模型构建问题,基于空间殖民算法提取树骨架,并提出树层次结构模型构建及曲率精简骨架的方法。空间殖民算法通过调整影响半径、删除阈值、骨架点距离和生长角度4个参数迭代生成树骨架;在设计树骨架、树枝干数据结构的基础上,依据树骨架点的关系构建树枝模型与树层次结构模型;为保留树枝弯曲特性,利用相邻骨架点曲率对骨架进行分类与精简;基于管道模型估算树枝干粗度,通过广义圆柱体渲染最终的树几何模型。重建结果表明,基于树点云重构的树三维模型与原点云吻合度较高,树形态结构自然,且树冠细枝结构突出,符合真实树特征。(4)为实现真实感树木运动模拟,基于提取的树运动动态参数、真实点云重建的树模型,用简易模态法实现了基于数据驱动的树动画模拟。通过fBm噪声模型构建树运动风速场,并基于空气动力学模型计算风力;采用简化的弯曲梁模型表示弯曲树枝干结构,振动频率和阻尼比作为时域运动方程的参数引入各占优振动模态的运动方程,通过求解简易模态分析方法的时域方程和空域方程实现树枝形变弯曲角度计算;整合弯曲树枝各节段的运动得到弯曲树枝整体运动,整合各弯曲树枝运动得到树整体运动。模拟试验结果表明,本文实现的树动画模拟系统交互式操作风向、风速的同时,可渲染帧率约为25~30fps的流畅树运动动画,效率较高;树的运动形态随风力与风向的改变而变化,中间层次长树枝弹性形变明显,树冠细枝随父树枝运动而运动,模拟的树运动符合冬季无叶玉兰树的特征;此外,对均匀点云及地面激光点云树模型的运动模拟表明本文模拟系统具有普适性。
[Abstract]:Realistic tree animation is of great significance to increase the appeal and immersion of the game. In recent years, in order to improve the realistic sense of tree animation simulation, data driven tree animation simulation method has become a hot topic, and the current data driven method is expensive, the process is complex and the gap between the tree model and the real model is large. In view of the above problems, this paper uses cheap Kinect somatosensory equipment to study the method of tree dynamic parameters extraction based on motion video and depth data and the building method of tree 3D model based on point cloud. Then the tree dynamic parameters and real real tree model are input into the tree dynamics model, and the model is based on the model. The main research contents and conclusions of this paper are as follows: (1) build a test platform for real-time data acquisition by using Kinect V2. Control the Kinect equipment erected on the triangular scaffold by PC workstation, and synchronize the video stream of branch trunk motion video with marked point. The experiment shows that the frame rate of the motion video stream and the corresponding depth information stream can reach 30fps (frames per second), and the collection is carried out. The dynamic data meet the needs of subsequent research. (2) a method of obtaining the vibration frequency and damping ratio of branches based on the local relative deflection angle criterion is proposed to obtain the expensive and complicated process of the tree real dynamic parameters. This method is used to track the two dimensional trajectories of the movement of the branches of the branches by tracking the color characteristics of the branches of the branches. Based on the tracking results, the two-dimensional tree skeleton motion model is constructed, and the local deflection angle of the sub branches relative to the parent branch is calculated. The time domain vibration signals are converted to the frequency domain by the fast Fourier transform. The frequency of the corresponding vibration peaks of the first order vibration peaks of the frequency domain signals is the frequency of the proposed branch vibration, and the damping ratio and the first order vibration frequency amplitude are used. The damping ratio is calculated by the semi wave method. The results show that the vibration frequency of the tested trees is 2.4Hz and 2.0Hz respectively. The vibration frequency of the tree is the inherent characteristic of the tree and does not change with the change of the external force; and the damping and external force of the tree are obviously related to the branch structure of the tree in the process of movement. (3) the data drive is solved. The real tree model of dynamic tree animation is built. Based on the spatial colony algorithm, the tree skeleton is extracted and the tree hierarchy model is constructed and the curvature simplification method is proposed. The spatial colony algorithm generates the tree skeleton by 4 parameters, which are adjusting the influence radius, removing the threshold, the distance of the skeleton point and the growth angle, and designing the tree skeleton and the trunk number. Based on the structure, the branch model and the tree hierarchy model are constructed according to the relationship between the tree skeleton points. In order to preserve the bending characteristics of the branches, the curvature of the adjacent skeleton points is used to classify and simplify the skeleton, and the trunk roughness is estimated based on the pipeline model and the final tree geometry model is rendered through the generalized cylinder. The reconstruction results show that the tree is based on the tree. The three dimensional model of point cloud reconstruction is higher than the original point cloud kiss, the tree structure is natural, and the tree crown fine branch structure is prominent. (4) in order to realize the realistic tree motion simulation, the tree model based on the dynamic parameters of the extracted tree motion, the real point cloud reconstruction model, and the simple modal method to realize the tree animation model based on data driven. The wind velocity field of tree motion is constructed by the fBm noise model, and the wind is calculated based on the aerodynamics model. The simplified bending beam model is used to express the bending branch structure. The vibration frequency and damping ratio are used as the parameters of the time domain motion equation to introduce the motion square of each dominant vibration mode, and the time domain formula is solved by solving the simple modal analysis method. The bending angle of the branch is calculated by the equation of the range and the space domain, and the motion of the curved branch is obtained by integrating the motion of the flexural branches, and the whole motion of the tree is obtained by the integration of the curved branches. The simulation results show that the tree animation simulation system is interactive with the wind direction and the wind speed is at the same time, the rendering frame rate is about 25~30fps The fluency tree motion animation has high efficiency, the tree movement pattern changes with the change of wind force and wind direction, the elastic deformation of the long branch in the middle level is obvious, the tree crown fine branch moves with the parent branch, and the simulated tree motion is in accordance with the characteristics of no Ye Yulan tree in winter. In addition, the motion simulation table of uniform point cloud and ground laser point cloud tree model is also made. The simulation system of this article is universally applicable.
【学位授予单位】:西北农林科技大学
【学位级别】:硕士
【学位授予年份】:2017
【分类号】:TP391.41

【参考文献】

相关期刊论文 前10条

1 赵春江;陆声链;郭新宇;杜建军;温维亮;苗腾;;数字植物研究进展:植物形态结构三维数字化[J];中国农业科学;2015年17期

2 胡鹏程;郭焱;李保国;朱晋宇;马韫韬;;基于多视角立体视觉的植株三维重建与精度评估[J];农业工程学报;2015年11期

3 蒋艳娜;肖伯祥;郭新宇;杨宝祝;;植物建模与动画合成研究[J];系统仿真学报;2015年04期

4 蒋艳娜;肖伯祥;郭新宇;杨宝祝;;基于视频的植物动画合成方法[J];中国农业科技导报;2015年01期

5 胡少军;耿楠;张志毅;杨沛;何东健;;基于稀疏图像的真实树交互式建模方法[J];农业工程学报;2014年09期

6 葛娟;李红军;张晓鹏;李华;;一个自适应的植物枝干重建方法[J];系统仿真学报;2013年05期

7 向泽锐;支锦亦;徐伯初;李娟;;运动捕捉技术及其应用研究综述[J];计算机应用研究;2013年08期

8 肖伯祥;郭新宇;陆声链;温维亮;郭小东;;植物三维形态虚拟仿真技术体系研究[J];应用基础与工程科学学报;2012年04期

9 唐卫东;胡雪华;郭晨;刘昌鑫;卢章平;;基于视频特征提取的虚拟植物动态仿真研究[J];计算机应用研究;2011年01期

10 周骏;陈雷霆;刘启和;李艳梅;饶云波;;树的建模技术研究综述与展望[J];计算机科学;2010年06期

相关博士学位论文 前1条

1 贺磊盈;面向振动采收的果树枝干三维重建方法及其动力学特性研究[D];浙江理工大学;2014年

相关硕士学位论文 前5条

1 路平;核桃采摘振动参数的研究[D];陕西科技大学;2014年

2 赵鹏飞;基于IFS的树木摇曳的动态模拟方法研究[D];西北农林科技大学;2013年

3 刘运婷;树在风中摆动的可视化模拟方法研究[D];西安科技大学;2010年

4 陆铖;基于视频的动物行为智能分析系统关键技术研究[D];西北农林科技大学;2008年

5 罗胜华;树随风动的计算机模拟研究[D];西安科技大学;2008年



本文编号:2082190

资料下载
论文发表

本文链接:https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2082190.html


Copyright(c)文论论文网All Rights Reserved | 网站地图 |

版权申明:资料由用户1c0db***提供,本站仅收录摘要或目录,作者需要删除请E-mail邮箱bigeng88@qq.com