融合高密度干涉条纹图谱和三维实物打印的微尺度显微立体视觉形貌测量法机理研究

Quantitative microscopic stereovision system which is constructed based on stereo light microscope has been applied in microscopic fields, such as microscopic measurement, micromanipulation, etc. And it is generally grown into a non-contact optical measuring method. It is mentioned that this method is commonly used to obtain geometry information of small-scale objects with sub-millimeter or several ten microns orders, and the disparity data is only derived from a stereo matching procedure based on initial stereo images. The microscopic measuring methods based on microscopic stereovision system related to stereo light microscope has many disadvantages in application, such as excessively depending on stereo matching from initial stereo images, poor stability and adaptability of matching results, and commonly containing massive wrong matching data, etc. Furthermore, the description methods for measuring results is only limited to numbers, characters and graphics, etc. In order to overcome these disadvantages, we focus on the study of surface profile reconstruction of micro-scale (several-micron and sub-micron orders) objects, which is carried out based on micro stereovision, interference fringe projection with high density and three dimensional printing techniques. A precision platform for measuring surface profile of micro-scale objects will be developed which combines microscopic vision, interference and three dimensional printing. And the detection of high density fringe map, the mechanism of micro-scale vision modeling, the mosaic of multiple-layer cloud data and the digital modeling of three dimensional printing will be studied deeply. At last, an efficient measuring method for profile of micro-scale objects is proposed based on the combination of microscopic vision, interference fringes and three dimensional printing, which can provide the theoretical and technical support for the development of microscopic stereovision technique with respect to stereo light microscope, and the function extension of corresponding measuring methods.

基于体视显微镜构建的显微视觉系统已在微测量、微操作等微观领域中使用，正逐渐成为一种非接触式的光学测量技术。当前，它多用于亚毫米或几十微米量级物体的几何信息测量，多依赖原始立体图像的匹配计算视差，存在匹配精度低、稳定性差、适应能力弱、容易出现大量错匹配数据等诸多问题，在测量模式上多以数字、文字、图形等常规展现方式为主。本项目申请针对上述问题，把已有显微视觉研究拓展到微尺度(几微米和亚微米)领域，以微尺度物体形貌重构为研究对象，以显微立体视觉、高密度干涉条纹投影、三维实物打印为技术手段，构建视觉、条纹投影、三维打印于一体的微尺度形貌精密测量平台，研究高密度条纹图谱的检测方法、微尺度视觉建模的机理、多层数据点云的拼接方法及三维打印数字化建模和工艺，面向微尺度物体形貌测量提出一整套融合高密度条纹图谱和三维实物打印的视觉测量新方法，为推动该类视觉技术的发展、拓展新领域提供理论和技术支撑。

该项目以体视显微镜双目立体视觉系统为基础平台，融合了属性可编辑的结构光投影系统，基于此研究结构光双目立体视觉技术和微小尺度物体的三维形貌重构方法，主要研究内容包括四个方面：形貌重构显微立体视觉测量系统的设计；显微条纹图谱视差曲面估计;结构光显微立体视觉数据重构;数据点云拼接和三维实物打印机理。在项目研究过程中，取得了如下的研究结果：(1) 建立了一种适用于量化显微双目立体视觉系统物像映射的方法，这种方法可以提供高精度的双目显微立体视觉系统的物像映射模型，模型的稳定性、精度和使用的便捷性明显优于已有的畸变矫正类的双目立体视觉模型；(2) 双目显微立体视觉系统与结构光能够结合组成主动式的显微立体视觉系统，可以更高效、更准确的获得小尺度物体表面的3D形貌，这种形貌重构原理在准确度和数据完整性方面具有优势；(3) 在结构光投影系统研制方面，使用投影仪产生结构光图谱，这种方法的适应能力较强，可以根据被测物体的表面光学属性来调整结构光的图案或颜色，以产生更好的结构光图像采集效果和重构质量。在机理研究的基础上，开展了大量的实验，对提出的显微双目立体视觉系统量化方法进行了充分的测试和验证，获得了较好的实验结果。已有的双目立体视觉系统量化方法多基于线性建模和畸变非线性校正的模式，用于微小尺度下的显微测量环境中容易产生局部计算结果不稳定的现象，计算结果的精度一致性较弱，本项目建立的显微双目立体视觉量化方法可以有效克服已有方法的不足, 对提升双目显微立体视觉原理的研究具有很好的推进意义，也有非常好的技术应用价值。