基于显微视觉的表面瑕疵检测与振动测量研究

CASIA OpenIR > 毕业生 > 博士学位论文

	基于显微视觉的表面瑕疵检测与振动测量研究
	陶显
	2016-05
学位类型	工学博士
中文摘要	随着我国大型装备制造和战略科学研究的需要，大口径光学元件在各种大型光学系统中应用越来越广泛，例如大型天文望远镜镜片、惯性约束核聚变系统中激光驱动器、资源卫星大视场角观测相机和导弹的光学成像导引系统等。由于大口径光学元件的应用场合特殊，其表面十分细微的瑕疵都会对整个光学系统产生极大的影响，因此对大口径光学元件的表面瑕疵进行准确和客观的检测具有重大意义。在工业场合中普遍存在许多柔性结构，这些柔性结构容易受到激励作用产生振动，进而导致整个结构出现损坏，因此对柔性结构的振动测量具有重大价值。目前，基于视觉的检测与测量方式在工业领域应用广泛，许多研究人员提出了各种针对表面瑕疵检测与振动测量的系统以及相应的算法，但是在实际应用中，仍存在问题有待解决。本文围绕显微视觉下的大口径光学元件表面瑕疵检测和柔性结构振动测量展开研究，主要的工作和贡献有： (1)本文设计了一套明场和暗场相结合的大口径光学元件表面瑕疵检测系统，暗场采用线阵相机搭配线光源的方式对瑕疵进行快速扫描，明场采用显微面阵相机搭配同轴光源对暗场检测的瑕疵进行进一步精确检测和测量，融合明场和暗场下的检测数据，得到最终光学元件表面瑕疵完整的检测结果。检测系统分别从图像采集子系统、运动子系统和软件子系统三个方面进行了详细设计。实验表明本系统在适应性、检测效率和检测精度等三个方面，与现有的检测系统相比有很大的提高。 (2)针对大口径光学元件的自身特点导致线阵相机扫描时容易离焦问题，提出了基于自动聚焦的线阵相机扫描路径规划方法，实现了大口径平面光学元件和非球面元件表面暗场图像的采集。针对于暗场图像的拼接问题，设计了基于SURF特征的图像拼接算法，实现了扫描暗场图像的快速拼接。针对大口径光学元件下多图像数据的显示问题，设计了基于图像金字塔的暗场图像显示方法，实现了暗场图像的高效和快速显示。针对暗场下散射效应影响瑕疵尺寸的问题，提出了基于明场和暗场融合的方式实现了瑕疵尺寸的准确测量。对上述方法进行了实验，实验验证了本文方法的准确性和有效性。 (3)为解决纳米级深度的划痕漏检和只检出部分长度等问题，提出了一种粗精结合的划痕检测算法，基于Gabor滤波器设计了线段探测算法对划痕进行粗检测，并采用基于LSD算法的划痕定位连接方法实现划痕的精检测。实验表明，该算法能够在低信噪比和边缘模糊情况下精确地检测出光学元件表面上的各种类型的划痕。提出了基于Kitter特征的瑕疵目标分割算法，实现了明场中低对比度的瑕疵目标的准确分割。针对瑕疵中损伤和污染的分类问题，设计了基于Gist特征分类方法实现划痕与污染的准确分类，提出了基于明场图像瑕疵纹理特征对灰尘和麻点进行准确区分的方法。 (4)针对柔性微小目标的振动测量问题，搭建了一套基于显微视觉的振动视觉测量系统，采用主动运动方式对图像雅可比矩阵进行标定，提出了采用圆环标志物和无标志物两种情况下的图像特征提取方法，利用图像雅可比矩阵对柔性目标的振动进行了精确测量。振动测量实验和对比实验验证了本文方法的有效性。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	With the need of large scale equipment manufacturing and strategic scientific research, large aperture optical elements are widely used in various large optical systems, such as large telescope lens, ICF laser driver system, satellite wide viewing angle observation camera and optical imaging missile guidance systems. Due to the application of large aperture optical elements, very small defects in the surface of the optical elements will have a great impact on the entire optical system. So it is of great significance to detect the surface defects of large aperture optical elements accurately and objectively. There are many flexible structures in the industrial field. These flexible structures are prone to vibration, which leads to the damage of the whole structure. So it is significant to measure the vibration of flexible structures. At present, the method of detection and measurement based on vision is widely used in the field of industry. Many researchers have proposed a variety of systems and the corresponding algorithms for surface defect detection and vibration measurement. But there are still problems to be solved in practical application. This dissertation focuses on the research on the surface defects detection of large aperture optics and the vibration measurement of flexible structure based on microscopic vision. The main work and contributions are as follow: (1) A surface flaw detection system for large aperture optical elements is designed, which combines the dark-field imaging system (DFIS) and the bright-field imaging system (BFIS). A line-scan camera with line source is used in the DFIS to the rapid scanning of the defect. An array camera with a coaxial light source is used in the BFIS, which carries on the further accurate detection and the measurement to the flaw. The final flaw detection results are obtained by fusing the detection data under bright and dark field. The detection system is designed in detail from three aspects: the image acquisition subsystem, the motion subsystem and the software subsystem. The experiments show that the proposed system’s performance is greatly improved compared with the existing detection systems’ in the three aspects of adaptability, efficiency and accuracy of detection. (2) Due to its own characteristics of large aperture optical elements, the line-scan camera is easy to out of focus in the process of scanning. In this work, the method of scanning path planning of line scan camera is presented for dark field image acquisition of the large aperture optical elements and aspheric. Aiming at the problem of dark field image stitching, an image stitching algorithm based on SURF features is designed to achieve a fast stitching of dark field scanning image. For the display of large image data, a display method based on image pyramid is presented to achieve efficient and rapid display of dark field images. For dark field scattering effect of defect size, the method based on the fusion of bright and dark field is proposed to realize the accurate measurement of the defect size. Experiments on the above methods have been carried out. The results show that the method is accurate and effective. (3) To solve the problems of nanometer depth of scratch detection and full length detection, this dissertation proposes an algorithm combining the coarse and fine scratches detection. A line segment detection algorithm based on Gabor filter is designed to detect the fine scratch. And the scratch location connection method based on LSD algorithm is used to realize fine detection. Experiments show that the algorithm can accurately detect various types of scratches on the surface of optical elements in the low signal-to-noise ratio and edge blur. In this dissertation, the defect target segmentation algorithm based on the characteristics of Kitter is proposed to realize the accurate segmentation of the defects in the middle and low contrast of the bright field image. The accurate classification method based on Gist feature for scratch and pollution is designed to classify damage and pollution defects. The texture features under bright field are used to accurately distinguish dust and pits. (4) Aiming at the vibration measurement of the flexible micro target, a vibration measurement system based on microscopic vision is set up. The image Jacobian matrix is calibrated by the active motion. In this dissertation, two kinds of image feature extraction methods are proposed based on two cases, i.e., circular ring marker and no marker. The vibration of flexible object is measured accurately by using the image Jacobian matrix. Experiments verify the effectiveness of the proposed methods. Finally, the results of the research are summarized, and the future work is analyzed and prospected.
关键词	大口径光学元件表面瑕疵检测振动测量显微视觉
学科领域	自动化
语种	中文
文献类型	学位论文
条目标识符	http://ir.ia.ac.cn/handle/173211/11533
专题	毕业生_博士学位论文
作者单位	中国科学院自动化研究所精密感知与控制研究中心
第一作者单位	精密感知与控制研究中心
推荐引用方式 GB/T 7714	陶显. 基于显微视觉的表面瑕疵检测与振动测量研究[D]. 北京. 中国科学院研究生院,2016.

条目包含的文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
陶显_毕业论文.pdf（10296KB）	学位论文		限制开放	CC BY-NC-SA