CASIA OpenIR  > 毕业生  > 硕士学位论文
分支结构的建模和采样
其他题名Modelling and Sampling Rami?ed Objects
尹巍巍
学位类型工学硕士
导师胡包钢
2005-05-01
学位授予单位中国科学院研究生院
学位授予地点中国科学院自动化研究所
学位专业模式识别与智能系统
关键词分支结构 隐式建模 点采样 子结构 等级结构 Ramified Structure Implicitly Modelling Point Sampling Substruc-ture Hierarchical Structure
摘要具有分支结构的物体是自然界普遍可见的物体之一,由于其几何结构和拓扑结构的复杂性,具有分支结构的物体一直是计算机图形学领域研究的热点和难点之一。传统的建模和可视化方法主要集中在用参数曲面和隐式曲面两种,前者注重快速的全局绘制,忽略了分支处的细节;后者虽然能很好地表现分支结构的局部细节,却不易快速绘制。本文针对分支结构建模和可视化进行了全面的研究,在采用隐式曲面的方法作为分支结构建模工具时,提出了对建模的隐式曲面进行点采样的方法,兼顾考虑了分支结构建模过程中的局部细节和快速绘制的两方面问题。本文的主要工作和贡献有:1° 针对分支物体结构复杂、具有一定自相似度的特点,提出了一种基于子结构的点采样方法。通过将分支系统分解为一系列具有等级的子结构,继而简化为基本单枝的线性组合,降低了系统结构的复杂度。同时,在底层创建了一个基本单枝的采样点库,通过高层的复制、变换等操作,直接获取整个系统的采样点。该方法通过缩减底层的直接采样目标,避免了对整个分支结构进行采样而产生的速度慢、参数难调等问题,提高了点采样的速度,同时充分利用已采样的子结构,避免对与其相同或者相似的结构进行采样的重复工作,从而进一步提到了点采样算法的效率。2° 利用隐式曲面易融合的特性,对分支区域内的分支粒子提出进行融合操作的方法,分别对椭圆融合、缩减骨架、超椭圆融合等方法进行了实现和深入的分析比较,最后采取了修正的超椭圆融合方法,通过分辩两骨架相交所形成的不同“凹”、“凸”区域,分别对分支粒子采取不同的融合方法,使分支粒子最终可以重新移动到分支之间光滑连接的融合曲面上,同时不引起额外的“膨胀”部分。3° 研究并实现了一个基于子结构的点采样软件,该软件可以根据输入的分支结构的骨架文件,快速获得融合后的整个分支系统的采样点信息并给予显示。总的说来,本文在针对一类特定(基于骨架、具有等级)的分支结构的建模和点采样工作中做了一些有益的探索。
其他摘要Ramified objects, one kind of ubiquitous objects, compose a great challenge in computer graphics due to their complex topological and geometric structures. Traditional methods for modeling and visualizing ramified objects can mainly be divided into two kinds. One is based on the parametric surface, which focuses on the fast global modeling and visualization while doesn’t care too much on local details in ramification. The other is based on the implicit surface, which shows unmatched advantages in generating smooth blending surface while have difficulty in real-time rendering. Through general research for modeling and visualizing ramified objects, both considering the local details in ramification and the fast global rendering, we adopt an implicitly modeling method and propose a point sampling algorithm. The main works and contributions of this thesis are listed as following: 1. Due to ramified objects having complex structures and self-similarity in certain degree, we propose a substructure-based point sampling method. By recursively dividing the full system into a set of substructures with certain orders and finally into the linear combination of basic branches, the structure complexity has been reduced. Meanwhile, we build up a sampling point library for basic branches in the low level, and get the sampling points for the full system by copying, transforming operations in the high level. Since this method shrinks and simplifies the sampling target in low level, it speeds up the sampling process and solves the problem of choosing parameters when we directly sample the full system. Additionally, the application of substructures helps to avoid the repetitious works for modeling and sampling the same or similar substructures, so it further improves the efficiency of sampling algorithm. 2. Utilizing the blending property of implicit surface, we propose to apply a merging operation to those particles that come from different substructures and locate in ramifications. We respectively implement the elliptic-merging, skeleton-reduction and super-elliptic merging operations and do a deep analysis and comparison. Finally, we suggest a modified super-elliptic merging operation, which separate the different regions that the two connected branches caused and applied different merging operations to the particles locating in different regions. After merging operation, all these particles have moved to the smooth blending surface and no unexpected bulge will exist. 3. We develop the software for our substructure-based point sampling method. According to different branching structure inputs, we can acquire all the sampling points for the merged surface fast and efficiently.
馆藏号XWLW859
其他标识符200228014603569
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/6895
专题毕业生_硕士学位论文
推荐引用方式
GB/T 7714
尹巍巍. 分支结构的建模和采样[D]. 中国科学院自动化研究所. 中国科学院研究生院,2005.
条目包含的文件
条目无相关文件。
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[尹巍巍]的文章
百度学术
百度学术中相似的文章
[尹巍巍]的文章
必应学术
必应学术中相似的文章
[尹巍巍]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。