【作者】 姜太平；

【导师】 沈春林；

【作者基本信息】 南京航空航天大学 ， 导航、制导与控制， 2007， 博士

【摘要】 在科学研究、仿真训练、游戏娱乐等方面我们经常需要显示三维图象。由于计算机屏幕是二维平面,我们从计算机屏幕上所见的只是用二维表示的三维图象。同时,围绕平面显示器,基于计算机图形学和数字图象处理技术得到广泛应用。但是这些三维图象缺乏物理深度,且不能同时由多人从多个不同角度去观察同一场景。基于此,本文研究了一种新型的立体显示技术-真三维立体显示技术。文中首先详细分析了体素的各种属性后,提出了一种TVF的真三维数据格式,对现有一些三维数据格式与TVF格式之间的转换进行了分析和实现,并在最后的仿真系统中进行了验证。然后,在对真三维立体显示系统的成像原理进行分类和分析的基础上,定量分析了有关成像过程,并在此基础上,对成像过程中的死区问题进行了深入探讨。其中重点分析了静态体成像过程中的死区问题。本文的研究重点是真三维立体显示系统的图象引擎设计。不仅提出了真三维图象引擎的系统构成,并就其中的体素激活技术,设计了有针对性的MC算法,获得了真三维立体场景的等值面。在此基础上,在借鉴二维显示技术的光线投影算法的基础上,结合TVF文件格式,提出一种新的体素预处理算法―发散式立体光线投影法(Radiate Volume Ray Projecting: RVRP)。该算法重点解决了图象引擎设计中的体素预处理问题,大大降低了真三维立体显示系统中的所需要激活的体素量,提高了整个真三维立体显示的速度,从而有效提高了真三维立体显示效果。当三维数据经过RVRP算法处理后,能够降低激活的体素数量,但是在体素激活的过程中,体素激活的顺序也会极大影响真三维立体显示效果。因此,本文对该过程进行了充分的数学分析,将该问题最终转换为传统的TSP(Traveling Salesman Problem:旅行商)问题,这是一个典型的NP-Hard问题。因此,本文研究了如何获得体素激活顺序的近似解。文中结合了体素激活的特征,提出了一种解决思路―“嵌套分块,路径优化,旋转扫描”。在嵌套分块中,提出了基于递归算法的循环嵌套分块算法(Recursive Nested Dividing Arithmetic : RNDA),可以根据不同的体素分布情况,快速有效地进行体素分块,降低每个分块中的体素激活量,从而降低了体素优化过程的计算量,并在基本遗传算法的基础上对其做了路径优化,使得该TSP问题最终可获得近似解。最后分析了立体显示技术中的自动立体三维显示技术的成像原理,即基于立体光栅成像法的显示技术和基于凸透镜成像法的显示技术。其中重点分析了自动立体三维显示器的多视组合算法,并作了软件和硬件方面的优化。在此基础上设计和开发了一套演示系统和二套飞行仿真程序,对本文中的部分研究内容和算法进行了验证。更多还原

【Abstract】 In the research of the simulating train,the game and the entertainment, the 3D images are necessary, but the computer display is the plate screen from which we only can get the dummy 3D images. Meanwhile, the various applications have been used with the technique of the computer graphics theory and the digital image processing. This kind of 3D effect lacks the physical depth and can’t support the several viewers’watching the same scene from the several angle of view. Then this dissertation researches a new volumetric display technology- True 3D volumetric display.Firstly, after the voxel attributes are analyzed, a new kind of true-3D file format(TVF) is put forward which can support the translation between the current 3D format and TVF format. This format is verified in the simulation system.Secondly, this dissertation describes the true 3D imaging process after classifying the theory of the true 3D volumetric display system. Then the dead zone of the imaging process is discussed deeply, and the dead zone of the static-swept imaging process is analyzed by mathematics.The main content of this dissertation is about designing the image engine of true 3D volumetric display system. It designs the composing of this engine, and put a new voxel pre-processing arithmetic referring to the ray tracing arithmetic of the 2D display. This arithmetic can reduce the amount of the voxel activation and improve the 3D image display’s speed and produce the better true 3D effects.After the voxels are processed by the RVRP mathematics, the activating amounts can decrease and simultaneity the activating order will affect the true 3D effects seriously. Then the dissertation analyzes the activating process, and translates it into the TSP by mathematics, which is the typical NP-Hard problem. To avoid such difficulty, how to get the similar answer is researched. It creates a new method-" Nested Blocking, order optimizing, rotary scanning" and also provides a recursive nested dividing arithmetic(RNDA). This arithmetic can divide the voxel rapidly and effectively by the voxel distributing. Then the calculating size can be reduces quickly.Finally the autostereoscopic display technique’s imaging theory and the the N-views combined arithmetic are analyzed which includes two kinds of the display techniques which base on the parallax or the convex. Then by programming a demo system and a simulation system of multi-angles flight onto a kind of autostereoscopic display, some of the researches are verified.The research of this paper commences the field of true 3D volumetric display technology domestically, and provides some theories and ideals for the later research and applications更多还原