作者：晏晖, 胡丙华

作者单位：中国飞行试验研究院, 陕西 西安 710089

关键词：空间拓扑关系;目标识别跟踪;位姿测量;影像测量;点云块配准;点距判定

摘要：

针对试飞光电测试中目标运动轨迹姿态自动化测量的问题，提出一种基于空间拓扑关系的目标自动识别跟踪与位姿参数同步测量技术。首先通过对被测目标表面的圆形标志点密集布设、目标运动过程的图像高速采集、圆形标志点不同视角下的自动精确提取，实现目标的自动检测；之后基于多点空间拓扑关系进行同名像对的匹配，并结合双摄像机的标定参数进行前方交会测量，实现目标表面标志点的三维重建；再采用基于多点空间拓扑约束、点距判定以及坐标旋转平移变换算法的点云块配准方法，实现目标自动跟踪与运动过程的轨迹姿态参数测量。试验证明：该方法切实可行，可获得亚毫米级的定位精度和优于0.11°的定姿精度，且能够全自动处理，从而大大提升试飞测试中影像测量的数据处理效率，同时可为飞行影像数据自动化、实时化处理提供一种有效技术手段。

Target automatic tracking and pose measurement technology based on spatial topological relations
YAN Hui, HU Binghua
Chinese Flight Test Establishment, Xi'an 710089, China
Abstract: Aiming at the problem of automatic measurement of target trajectory and attitude in flight test, an automatic target recognition tracking and pose synchronous measurement technology based on spatial topological relations is proposed. Firstly, the target is automatically detected by densely arranging circular marking points on the surface of the target, high-speed image acquisition of the target motion process, and automatic and accurate extraction of the circular mark points at different viewing angles. After that, image matching of homonymous image pairs based on the multi-point spatial topological relations and is performed, which combining the calibration parameters of the two cameras, the forward intersection measurement is used to achieve the three-dimensional reconstruction of the target. And then, point cloud registration method combined with multi-point spatial topological constraints, point distance determination, and coordinate rotation translation transformation algorithm is used to achieve target automatic tracking and pose measurement. Experiments have proved that this method is feasible and can obtain sub-millimeter-level positioning accuracy and positional accuracy better than 0.11°, and it can be fully automated, which greatly improves the data processing efficiency of image measurement in flight test and provides an effective technical means for automated and real-time processing of flight image data.
Keywords: spatial topological relation;target identification and tracking;pose measurement;image measurement;point cloud registration;point distance determination
2019, 45(4):9-15  收稿日期: 2018-05-14;收到修改稿日期: 2018-06-28
基金项目: 装备预研联合基金（614B05110302）；航空基金（2015ZD30002）
作者简介: 晏晖(1984-),男,陕西汉中市人,高级工程师,硕士,主要从事摄影测量研究与应用以及光电测试工作
参考文献
[1] YAN W, FENG Y, HONG J, et al. High precision and fast estimation of position and attitude measurement for space targets[J]. Optik-International Journal for Light and Electron Optics, 2017, 148:76-84
[2] LIU W, MA X, JIA Z, et al. Position and attitude measurement of high-speed isolates for hypersonic facilities[J]. Measurement, 2015, 62:63-73
[3] 孙鹏, 吕乃光, 王博恩, 等. 视觉测量中圆形标志点的全场自动识别和提取[J]. 北京信息科技大学学报, 2011, 26(3):91-95
[4] ROCIO G, MARIA-JOSE J, BELEN M. Topological tracking of connected components in image sequences[J]. Journal of Computer and System Sciences, 2018, 95:134-142
[5] KRISTOFFER S, ALPER A, PATRIC J. Topological spatial relations for active visual search[J]. Robotics and Autonomous Systems, 2012, 60:1093-1107
[6] 金兆楠. Zernike矩的圆阵列靶标特征点精确提取方法[J]. 山西电子技术, 2015(4):37-39
[7] 张剑清, 潘励, 王树根. 数字摄影测量学[M]. 武汉:武汉大学出版社, 2006
[8] 张永军. 基于序列图像的视觉检测理论与方法[M]. 武汉:武汉大学出版社, 2008
[9] 聂建辉, 马孜, 胡英. 一种鲁棒的圆形标记点中心提取方法[J]. 应用科学学报, 2013, 31(1):91-96
[10] 赵洛彬, 杨瑞峰, 郭晨霞. 靶标特征点提取亚像素精度阈值分割算法研究[J]. 中国科技论文, 2015, 10(8):942-947
[11] CHEN Y, ANAND S, MARTINEZ-CONDE S, et al. The linearity and selectivity of neuronal responses in awake visual cortex[J]. Journal of Vision, 2009, 9(9):12-17
[12] 舒禹程. 基于特征描述子的图像匹配算法研究[D]. 武汉:华中科技大学, 2015
[13] 雷玉珍, 李中伟, 钟凯. 基于随机抽样一致算法的误匹配标志点校正方法[J]. 光学学报, 2013, 33(3):1-8
[14] 谢晓燕, 吴锦桥. 一种全自动三维点云配准及比例约束方法[J]. 计算机技术与发展, 2015, 25(3):63-66