复杂曲面高精高速非接触检测理论与精度增长技术研究

Complex surface parts are key components used in huge-size equipments,and its manufacturing quality is become more and more important for the equipments keeping a good running state and high reliability. Aiming to improving the measurement accuracy for inspection of complex surfaces in manufacturing workshop, the contribution of the project is described as followings. By studying the theories of optical methodology and data processing, and analyzing the generation, transmission, increasing and varying of measurement errors, several available methods to "see" and "control" the measurement errors are developed. On the basis of locating the main components lead to the measurement errors, the solutions of removing or decreasing the measurement errors are found out. A new modified projection fringe based method is proposed to remove the errors caused by fringe torsion in both fringe projection and CCD imaging stages. Some specialized image processing methods with the blur estimation, texture analysis are proposed to attack the difficulties of segmentation of the blur images generated from defocus fringe projection. In addition,a new method is proposed for functional approximation of data points using MLS (Moving Least Squares)-based surfaces so as to decrease the stochastic errors of inspecting process resulted from noise, environment light,and material properity, etc for improving the accouracy of data processing. It is desired that a fringe projection-based system could be developed for industrial inspection with high speed and accuracy and reverse engineering by using proposed methods.

复杂曲面零件是大型机械装备的核心部件，为了保证系统工作过程的可靠性与稳定性，对其制造精度提出越来越高的要求。本项目以大型复杂曲面检测需求为背景、提高测量精度为目标，研究测量过程及数据处理的相关原理，揭示误差的形成、传递、积累与协调机制，探索测量过程误差的"可观"与"可控"机理，在搞清影响测量精度关键环节的基础上，挖掘消除误差的途径与技术。研究测量过程中条纹投影与CCD拍摄阶段的映射特征，揭示工作过程中误差产生与演变机制，提出通过修正条纹减少误差的方法；研究非焦点条纹投影测量的原理特征，揭示非焦点成像的机理与数据特点，提出通过建立调制函数、纹理模糊度度量与标定等途径实现精度增长；研究基于最小二乘法理论为基础的动态局部数据建模方法，揭示随机测量数据误差的传递规律和影响，提出一套满足复杂曲面检测要求的数据处理与精度保障技术，为提高复杂曲面的制造质量、测量误差评定与实物反求等提供理论与技术支撑。

现代工业设计中，复杂曲面零件的应用非常广泛，对零件的制造和检测精度要求越来越高。本项目以复杂曲面的非接触检测为背景，提出几种提高光学三维测量精度和速度的方法，并通过搭建实验平台验证方法的有效性。主要按以下几个方面研究论述：.1. 研究了条纹修正技术，通过分析条纹投影拍摄过程中的误差来源和传递过程，提出了基于反向投影的非均匀条纹生成算法；然后建立条纹灰度评价模型，采用灰度迭代与消除退化项的处理实现条纹灰度的调整，提高中心提取精度；并提出一种快速准确的条纹匹配方法，实现三维点云计算。.2. 研究非焦点投影中的关键技术，针对模糊图像，提出了一种基于Mumford-Shah模型的角点特征提取算法，降低标定过程中的精度丢失；然后根据非焦点投影中的光学特征，提出了一种基于相位映射和投影仪平面镜头畸变矫正的二值散焦投影仪标定方法。.3. 研究多测头组合测量和局部数据处理技术，以测头的视场重叠区域为目标，提出了多测头系统中的测头配置和空间布局优化方法；然后通过自由组合双目结构光测量系统，提出基于移动最小二乘和最近点迭代的提高测量精度和数据可靠性的方法。.4. 建立了基于光学测量的在线检测系统，针对2.5D和回转体零件的零件特征，提出对应的测量方法；然后设计了一种基于激光传感器的同轴度检测装置，并提出一种针对三维点云数据的最小二乘迭代算法，实现同轴度测量。