【作者】 金鹭；

【导师】 杨将新； 曹衍龙；

【作者基本信息】 浙江大学 ， 机械制造及其自动化， 2011， 硕士

【摘要】 本文结合国家自然科学基金重点资助项目“复杂机电产品质量特性多尺度耦合理论与预防性控制技术”(50835008)和国家科技重大专项项目“基于敏感点监测的闭环动态综合补偿技术”(2009ZX04014-026),对三维表面形貌测量及其质量评定、基于图形处理器(Graphics Processing Unit, GPU)的并行处理算法等进行了研究,提出了“基于GPU的表面形貌测量系统研究”的硕士论文研究课题。第一章：阐述了表面形貌测量的背景和意义,介绍了国内外表面形貌测量及其质量评定的发展和现状,阐述了基于图形处理器GPU的并行处理技术的研究现状,提出了论文研究背景和论文框架。第二章：阐述了用于表面形貌测量的白光谱线扫描干涉法的测量原理,并分析了相应的相位计算算法。通过对比分析,选择了波峰计数法和傅里叶变换分析算法作为干涉图像处理算法。第三章：论述了表面形貌测量的数据处理过程,首先分析了基于统一计算设备架构(Compute Unified Device Architecture, CUDA)的GPU并行算法的设计方法。其次分析了信号数据预处理的方法,基于GPU分别设计实现了波峰计数法和傅里叶变换分析法两种相位并行计算方法,并对比分析了两种并行算法的处理速度和处理精度。最后在实验对比分析的基础上,确定了傅里叶变换分析法作为基于GPU的表面形貌测量系统的相位计算算法。第四章：研究了高斯滤波基准面的建立方法,使用GPU对二维高斯滤波过程进行并行处理,并在此基础上给出了三维表面粗糙度评定并行处理过程,最后对标准表面粗糙度比较样块进行了测量和结果分析。第五章：搭建了使用白光扫描干涉测量法的表面形貌快速测量系统。设计了系统的控制电路和测量系统软件的各个功能模块,完成了整套系统的开发,对表面三维粗糙度进行了评定。第六章：概括全文内容,展望了基于GPU的表面形貌测量系统的发展前景。更多还原

【Abstract】 Based on the National Natural Science Foundation of China (No.50835008) and the important national science & technology specific projects (No.2009ZX 04014-026), surface topography measurement and surface topography evaluation and the application of Graphics Processing Unit (GPU) are studied.In chapter one:The importance and background of surface topography measurement are expounded. Introduced the development and current situation of surface topography measurement and evaluation, expounded the development of scientific computing by using GPU. Finally, the study background and overall frame of the dissertation are given.In chapter two:White-light spectral scanning interferometry has been proposed. Several phase calculation algorithms are introduced. Through the comparative analysis, choose peak counting algorithm and Fourier transform analysis algorithm for Interference image processing algorithms.In chapter three:The data processing of surface topography measurement is discussed. First of all, the design of GPU parallel algorithm based Compute Unified Device Architecture (CUDA) is discussed. Secondly, analyzed preprocessing methods of signal, then realize parallel algorithm of peak counting and Fourier transform analysis by using GPU. The processing speed and the precision of these two parallel algorithms are analyzed. Finally, through the experiences, the Fourier transform analysis algorithm is determined as the phase calculation algorithm in surface topography measurement based on GPU.In chapter four:The establish method of three-dimensional Gaussian filter datum is studied, use GPU to processing two-dimensional Gaussian filtering procedure, and then propose the parallel processing method of evaluation of three-dimensional surface roughness. At last, standard surface roughness comparison specimens are measured and the results are analyzed.In chapter five:Based on the theory study, a surface topography measurement system is developed by using CUDA. The system control circuit and software module structure are designed, and the development of whole system is realized, and evaluated the surface roughness.In chapter six:The main content of the dissertation research is summarized, and the prospects of surface topography measurement based on GPU are presented.更多还原