【作者】 张伟；

【导师】 胡国清；

【作者基本信息】 厦门大学 ， 机械制造及自动化， 2009， 硕士

【摘要】 随着计算机图形学和三维扫描技术的不断发展,点模型受到了越来越多的重视。由于数据点既包含了模型的几何信息(如空间坐标、法向量等),又包含了模型的表面属性(如颜色、光照等),所以大大简化了空间数据结构和算法。本论文对空间点云消噪进行研究。在实际采集数据过程中,由于数据采集环境和完成数据采集任务的仪器仪表自身的原因,不可避免地存在其它因素(如操作方法是否得当、环境的影响等)的干扰和噪声,所以在采集的数据点中会存在噪声。噪声点或是奇异点的存在对于数据采集和信号测量之后的科研和生产工作会造成不利的影响这些噪声信号或奇异点将掩盖、扭曲我们所需要的有用数据,因此在对数据进行处理之前必须对其进行噪声点的处理,以便消除数据的噪声点,对原始数据进行重构,有效地表现原信号中的有用的信息。重构的效果将直接影响到后续的基于信号故障识别、分析、诊断等方面的工作的进行。本论文是逆向工程的应用。在对点云数据进行消噪的过程中,引入了小波理论。小波理论从上世纪70年代诞生至今,不断地发展和完善。小波分析的应用是与小波分析的理论研究紧密地结合在一起的。目前,它已经在科技信息产业领域取得了令人瞩目的成就,如在工程技术等方面,其中包括计算机视觉、计算机图形学、曲线设计、湍流技术、远程宇宙的研究与生物医学方面。现今,计算机图形学和信号处理技术已经成为当代科学技术工作的重要部分,信号处理的目的就是:准确地分析、诊断、编码压缩和量化、快速传递或存储、精确地重构(恢复)。从数学角度来看,信号与计算机图像的处理可以统一看作是信号处理(图像可以看作是二维信号)。除此之外,小波分析的许多应用,也都可以归结为信号处理问题。本论文正是基于小波分析的种种优势,而将其引入点云数据的消噪中。本文的思路如下:1.通过一定的数据采集方法获取模型的空间点云数据。2.提出将空间点云片层化的思想,并对空间点云数据的分层方法进行讨论。通过比较几种常见的数据分层方法的优劣,并根据实际情况及要求,找到一条简单可行的途径,并对该途径的可行性进行理论上的分析与论证。3.将“切片”后的三维坐标降维,降至二维平面内,为后面用小波进行消噪处理打下基础。4.对小波理论进行深入的学习,主要包括多分辨分析理论的研究和利用小波进行奇异性检测的原理。5.通过理论的学习研究,对“切片”后的数据进行小波分解,检测并去除奇异点。6.对点云进行整合,重构实体模型。更多还原

【Abstract】 With the developing of computer graphics and scan technique,the model of data points is becoming more and more important.The geometry information and surface attribute are involved in the model of data points,so the database and method of computing are predigested primely.In the thesis,the denoising of data points is studied.In the process of collecting data points,owning to a few of unavoidable factors that comes from instruments and the environment,the errors are inevitable.The noise points which distort the true data will bring a lot of disadvantage to the work of data collecting and projecting.Therefore the noise points must be wiped off before the project.When the noise points are got rid of,the original model of data points,which contains useful information,can be constructed primely.The effect of data constructing will affect the subsequent work of identifying, analyzing and diagnosing.The task of this thesis is an application of reverse engineering.The wavelet theory is used in the process of the work of denoising.The wavelet theory was produced in 1970s. Since then this theory has been reformed continuously.The applications of wavelet are combined with the wavelet theory.Nowadays the applications of wavelet have taken tremendous success in the field of information industry.For instance,the field of engineering,which include computer vision,computer graphics,curve design,onflow technology,long-distance universe study,biomedicine and so on,has developed enormously.Especially,the computer graphics and signal processing have been already the important components of modern technology,the purpose of signal processing is analyzing accurately,diagnosing,coding,transferring rapidly,storage and resuming accurately.From the view of mathematics,the management of computer image and signal can boil down to signal processing(because the computer image can be seen as planar signal).Furthermore other applications of wavelet could be also considered to be signal processing.In view of a good many advantages,the wavelet method is used in this paper.The main study results of this thesis are summarized as follows:1.Obtaining the data points of an experimental model by a special method of collecting data.2.A delamination method is put forward,and the delamination methods of data points are discussed.A feasible delamination method is put forward by comparing several common ways,and the feasibility of the new method is demonstrated in this paper.3.Transform the three-dimensional coordinates to planar coordinates.This step is the precondition of denoise work.4.Study the wavelet theory and the multi-resolution analysis.5.Analyze the planar data with wavelet method that is learned at the fore.6.Construct the entity with the new points cloud.更多还原