基于显微视觉的微管-微球装配与胶接研究
文献类型：学位论文
| 作者 | 李福东 |
| 学位类别 | 工学博士 |
| 答辩日期 | 2014-05-23 |
| 授予单位 | 中国科学院大学 |
| 授予地点 | 中国科学院自动化研究所 |
| 导师 | 徐德 |
| 关键词 | 显微相机标定 自动聚焦 姿态测量 微胶接 微装配 microscopic camera calibration auto-focusing pose measurement micro adhesive bonding microassembly |
| 其他题名 | Research on Micro Tube and Micro Sphere Assembly Involving Adhesive Bonding based on Microscopic Vision |
| 学位专业 | 控制理论与控制工程 |
| 中文摘要 | 随着微机电系统功能的增多，传统的微加工技术已不能满足这种需求。将不同材料，不同加工工艺制造的微零件进行组装与连接从而形成功能强大的混合微机电系统已成为一种趋势。显微视觉下的微装配技术能有效地解决三维微机电系统装配问题。本文针对微装配中的显微图像处理、显微相机标定以及显微视觉下物体在三维空间中的姿态测量等显微视觉问题展开研究。主要的工作和贡献有： （1）提出了一种基于显微视觉的微管与微孔对准和插入控制方法。由于显微视觉景深小，无法使整个物体成清晰图像，并且物体在移动过程中会远离景深范围而造成图像模糊，目标丢失。针对这些问题，提出了一种控制微管末端在聚焦平面运动的策略，使其在运动过程中始终保持清晰。成功实现了微管自动对准与插入微孔。 （2）提出了一种稳定可靠的基于灰度统计与搜索框结合的微孔区域定位算法。由于半透明的微球在高倍率的显微相机中成像质量较差，而且微球的球形结构由于光照条件在显微相机中成像不均匀。这造成稳定可靠的微孔区域定位非常困难。根据微孔图像较暗，且比较孤立的特点，设计了一种首先根据灰度统计比例确定灰度阈值，然后设计搜索框进行微孔区域搜寻的算法。该算法能够稳定可靠地对微孔区域进行定位，对光照条件变化适应性好，而且对微孔离焦的情况也能准确地进行定位。 （3）提出了一种基于主动运动的显微相机标定方法。传统的相机标定需要用到标定块或者标定板，而由于显微相机的景深小、视野小，使得标定块与标定板的应用受到很大限制。基于主动运动的显微相机标定方法利用微管操作手的高运动精度，控制微管末端在显微相机的聚焦平面上运动两步。通过将微管末端的图像坐标与相应的微管操作手的电机坐标进行变换处理，对显微相机与微管操作手的手眼关系以及显微相机的内参数进行标定。 （4）提出了基于显微视觉的微管与微孔在三维空间中的姿态测量与对准方法。针对显微相机景深小的特点，通过显微相机的聚焦运动，分别测量出待测姿态向量与显微相机光轴平行线所构成的平面。通过计算左右显微相机中的两个平面的交线计算出待测姿态向量在微管操作手坐标系中的表达。并根据微孔与微管姿态相差不大的特点，提出一种通过绕x轴和y轴旋转微球夹持器来调整微孔姿态使之与微管姿态进行对准的方法。 另外，针对微管与微球胶接的胶斑控制，通过设定胶斑直径检测框，并在框内进行背景差分对胶斑的直径进行在线检测。有效地将胶斑直径控制在30~40um之间。 最后，总结了研究所取得的成果，并对后续工作进行了分析和展望。 |
| 英文摘要 | As MEMS (Micro-Electro-Mechanical Systems) is becoming more and more functional, conventional MEMS fabrication technologies are insufficient in producing MEMS with multiple functionalities. Integration of MEMS devices fabricated with different technologies, different materials is required to build multi-functional hybrid MEMS. Microscopic vision guided microassembly technology is effective in constructing hybrid MEMS. This paper is mainly about researches on microscopic image processing, microscopic camera calibration and pose measurement based on microscopic vision. Main work and major contributions are: (1) A control strategy for micro tube-hole alignment and insertion based on microscopic vision is presented. Because of the microscopic vision has small depth of field, clear image of the entire object can not be achieved. Target tracking will fail when the target is moved out of the focal plane and causing the image severely blurred. To solve these problems, a strategy to control the tip of the micro tube to move on the focal plane is developed, and the image of the tube’s tip will keep clear during the alignment. Automated aligning and inserting of the 10 m diameter micro tube into the 12 m diameter micro hole is realized. (2) A method to stably and reliably locate the hole area based on gray statistics and searching window is introduced. Stable and reliable localization of the hole is difficult to achieve due to the fact that the image quality of the semitransparent micro sphere under high magnification microscopic camera is very poor, and the lighting condition is nonuniform because of the spherical structure. A gray value threshold is determined by gray statistics according to the fact that the hole image is dark, then a searching window is used to locate the hole area because the hole image is isolated. The proposed method is very stable and reliable in locating the hole area, and it is robust to various lighting conditions and out of focus images. (3) A motion based microscopic camera calibration method is proposed. Traditional camera calibration methods need calibration patterns, but the usage of calibration patterns in the microscopic vision is limited by the small field of view and small depth of field of the microscopic vision. In the motion based microscopic camera calibration method, the tip of the micro tube is moved on the focal plane for two steps. With the processing of the recorded image coordinates of the micro tube... |
| 语种 | 中文 |
| 其他标识符 | 201118014628011 |
| 源URL | [http://ir.ia.ac.cn/handle/173211/6590]  |
| 专题 | 毕业生_博士学位论文 |
| 推荐引用方式 GB/T 7714 | 李福东. 基于显微视觉的微管-微球装配与胶接研究[D]. 中国科学院自动化研究所. 中国科学院大学. 2014. |
入库方式： OAI收割
来源：自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。