【作者】 李满海；

【导师】 陈文静；

【作者基本信息】 四川大学 ， 光学， 2004， 硕士

【摘要】 光学三维面形测量在工业测量、机器视觉、实物仿形、制鞋服装设计、生物医学等领域具有重要意义和广阔应用前景。基于条纹投影的傅里叶变换轮廓术(FTP)具有单帧获取，全场分析和高分辨率等优点，成为三维传感的基本方法之一。随着计算机技术硬件和软件的发展，以及图像获取设备分辨率的提高，傅里叶变换轮廓术倍受人们的关注，成为三维传感中最重要和最活跃的研究领域之一。 本文在离散傅里叶变换分析理论的基础上，深入讨论抽样对FTP的影响。重点讨论CCD的取样模型，致力于分析实际测量中CCD像元对FTP测量精度的影响，完善了FTP中的抽样模型。最后从频域滤波角度分析提高FTP测量精度的新方法。主要研究成果如下： 1、针对讨论抽样对FTP影响中存在的缺陷，着重研究了CCD成像所完成的第一次取样，以及CCD的取样中所涉及信号的带宽限制和频谱混叠问题。建立了CCD的取样模型，分析了CCD像元的尺寸对FTP测量精度的影响，并给出了相关的表达式和判断准则，使得FTP的研究更贴近实际测量。 2、由于频谱滤波是FTP中很重要的一个过程，也是影响FTP测量精度的一个重要因素。本文提出一种门限滤波方法，针对正弦结构光场情况，这种滤波方法无需手工选择滤波函数，又可以很好地保留有效高频信息，并有效消除CCD采集的变形光栅条纹中混入的随机干扰噪声。更多还原

【Abstract】 Optical 3-D profilometry is used widely for industry measurement, machine vision, profile modeling, shoemaking engineer, biomedical industry, etc. Fourier Transformation Profilometry (FTP) based on fringe projection is one of an important 3-D sensing method, including following merits: only one fringe needed, full field analysis, and high precision. With the development of computer hardware and software technique and high-resolution image grabber, FTP has become one of the most important and popular methods and payed more and more researchers’ attention.Based on the discrete Fourier transform analytical method, this paper discusses the influence of the pixel size of CCD and proposes a new filtering method. The main results can be summarized as following:1 Aimed at the problems existed in the study of influence of sampling in FTP, we discuss the first sampling when CCD captures the information of object., and discuss the influence of the pixel size of CCD, such as the limitation of bandwidth and the frequency overlapping. Now we discuss the sampling model of CCD and give the expression and sampling selection criteria. The discussion makes the research of sampling theory in FTP more actually and practically.2 Filtering is an important process in Fourier transform profilometry, and it is also an important factor for influencing precision. In this paper, a new filtering method is proposed, which is an efficient method to improve the measuring precision. This method can keep enough frequency and eliminate the random noise,and there is no need to choose a filtering window.更多还原