CORC  > 自动化研究所  > 中国科学院自动化研究所  > 毕业生  > 博士学位论文
题名基于视觉的靶精密定位与多光束引导技术研究
作者张鹏程
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
关键词显微视觉标定 靶的跟踪与定位 多光斑跟踪与引导 视觉反馈控制 束靶耦合 Microscopic Camera Calibration Target Tracking and Positioning Multiple Laser Beams Tracking and Guiding Visual Feedback Control Beam and Target Alignment
其他题名Research of Target Precision Positioning and Multiple Laser Beams Guiding Technologies Based on Vision System
学位专业控制理论与控制工程
中文摘要激光惯性约束核聚变能量密度高、反应高效,可提供丰富、经济和安全的能源,已成为解决未来能源危机的有效途径之一。束靶耦合是激光惯性约束核聚变的关键技术之一。在核聚变之前,首先,利用靶精密定位装置将目标靶精确定位到真空靶室中心附近一个很小的范围内,并建立靶室基准。然后,引导多束高功率激光在极短时间内聚焦到微型靶丸上。因此,如何完成目标靶的高精度跟踪和定位,以及如何引导多激光束实现对目标靶的精确瞄准已经成为束 靶耦合实验亟待解决的问题。 本论文在国家自然科学基金支持下,以束靶耦合实验平台为对象,就束靶耦合过程中的靶精密定位和多光束引导等关键技术展开研究。其内容涉及到显微视觉标定算法、靶精密定位和视觉控制算法、靶的三维跟踪和送靶控制算法、多光束跟踪和引导算法等方面。本文的主要研究工作和贡献可概括为: (1)提出了基于一维标定物的显微视觉系统内外参数的自标定算法。针对单目显微视觉景深小、视野小的问题,其成像模型和标定方法有别于宏观视觉,采用一维标定物生成的虚拟标定平面进行内参数标定,并通过对比基于弱投影模型法、简化的外参模型法、改进的Tsai法和改进的Zhang 法四种标定方法,选择最优的单目显微视觉内参数标定方法。实验结果表明,提出的标定方法的投影误差为0.8像素,反投影误差为19.2μm。针对多目显微视觉系统不存在公共视野的问题,通过清晰成像平面实现了多目显微视觉之间的位姿变换关 系的标定。 (2)提出了靶目标精密定位和视觉控制算法,并提出了基于图像雅可比矩阵或清晰成像平面的多目显微视觉信息融合技术。针对三目显微视觉下基于图像的靶精密定位问题,利用图像雅可比矩阵将图像空间像素偏差转化到笛卡尔空间的目标位姿偏差,设计控制器,使当前图像特征收敛到期望的图像位置。针对基于位置的靶精密定位问题,通过显微视觉的清晰成像平面将三目显微视觉子系统构成立体视觉系统,在笛卡尔空间设计控制器,使目标靶当前位姿收敛于期望位姿。通过实验结果和分析,选择针对束靶耦合实验平台的精定位方案,靶目标定位的位置精度可达10μm,姿态精度可达0.02◦。 (3)提出了基于Eye-to-Hand单目宏观视觉配置的送靶方法。针对目标靶位姿估计问题,研究了基于CAD模型的三维跟踪算法,利用虚拟视觉伺服技术,通过在图像空间中模型投影边缘收敛于当前帧的图像边缘实现在笛卡尔空间中CAD 模型与目标靶三维对准,实时获得目标靶的位姿。针对送靶控制问题,采用手眼标定技术,获得摄像机坐标系和机器人基坐标系,以及目标靶坐标系和机器人基坐标系的位姿变换关系,构建视觉反馈闭环系统,将靶定位于显微视觉视野范围内,完成送靶控制。实验表明送靶系统的位置误差控制在0.2mm 以内,姿态误差控制在0.15◦以内。 (4)提出了多光斑检测、跟踪和引导策略,通过构建视觉反馈系统,实现多光束的并行瞄准。针对多光斑跟踪问题,利用基于链码的轮廓跟踪算法实现多光斑跟踪,并通过图像矩实现各个光斑的形状识别。针对光斑混叠问题,利用卡尔曼滤波预测光斑位置坐标,并通过多光斑的相互遮挡率调节卡尔曼增益,实...
英文摘要Inertial confinement fusion (ICF) has the property of high energy density and can provide clean, economic and secure energy, which has become an effective way to solve the problem of future energy crisis. Beam-target alignment is one of the key technologies of ICF. First of all, before the fusion, the target is positioned precisely near the center of the vacuum target chamber and a high-accuracy coordinate system is established. Then multiple laser beams are guided to shoot the target in a short time. Therefore how to accomplish the task of high-precision target positioning and how to guide multiple laser beams to shoot the target accurately have become urgent for the beam-target alignment experiment.Supported by the National Nature Science Foundation of China, the precise and robust approaches for positioning target and guiding multiple laser beams are developed based on the beam-target alignment experimental platform. The research contents involve microscopic camera calibration, target positioning and visual control algorithm, target 3D tracking and sending techniques, multiple laser spots tracking and guiding strategies.The main work and contributions are described as follows, (1)A calibration technique for microscope vision based on one virtual pattern which is constructed by a calibration sphere is presented. Because of the small depth-of-field and field-of-view, the calibration method for microscope vision has unique characteristics which are quite different from macro-vision calibration, including the camera model and calibration pattern. A virtual pattern is constructed by a high-precision positioning robot and is used for calibration. The adequate calibration approach for microscope vision is selected by comparing four methods, including the weak perspective method, the simplified parameters method, the modified Tsai’s method, the modified Zhang’s method. Experimental results show the projection error is 0.8 pixels and the back projection error is 19.2μm for the proposed one. Because of the case that no common features exist in different microscope vision, the relationship between ones is calibrated by the corresponding clear imaging plane.(2)The precise positioning and tracking techniques are proposed and the fusion method of multiple vision sub-systems by the image Jacobian matrix or the clear imaging plane is proposed. For the target positioning task based on three microscope vision systems, an image-based visual control (IBVC) with imag...
语种中文
其他标识符201118014628024
内容类型学位论文
源URL[http://ir.ia.ac.cn/handle/173211/6591]  
专题毕业生_博士学位论文
推荐引用方式
GB/T 7714
张鹏程. 基于视觉的靶精密定位与多光束引导技术研究[D]. 中国科学院自动化研究所. 中国科学院大学. 2014.
个性服务
查看访问统计
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

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。


©版权所有 ©2017 CSpace - Powered by CSpace