基于单光子计数三维成像的目标姿态获取方法  

Single-Photon Counting 3D Imaging-Based Target Attitude Acquisition

作　　者：张瀚夫 刘杰[1,3,4] 安其昌 王建立[1,2,3,4] Zhang Hanfu;Liu Jie;An Qichang;Wang Jianli(Photoelectric Detection Technology Research Department,Changchun Institute of Optics,Fine Mechanics and Physics,Chinese Academy of Sciences,Changchun,130033,Jilin,China;University of Chinese Academy of Sciences,Beijing 100049,China;Jilin Key Laboratory of Intelligent Wavefront Sensing and Control,Changchun,130033,Jilin,China;Key Laboratory of Space Object and Debris Observation,Chinese Academy of Sciences,Nanjing 210008,Jiangsu,China)

机构地区：[1]中国科学院长春光学精密机械与物理研究所光电探测技术研究部,吉林长春130033 [2]中国科学院大学,北京100049 [3]吉林省智能波前传感与控制重点实验室,吉林长春130033 [4]中国科学院空间目标与碎片观测重点实验室,江苏南京210008

出　　处：《激光与光电子学进展》2023年第8期370-377,共8页Laser & Optoelectronics Progress

基　　金：国家自然科学基金(62135015,62005279);国家重点研发计划(2020YFA0714102);吉林省科技发展计划(20220402032GH);中国科学院青年创新促进会(2020221)。

摘　　要：单光子计数成像技术在暗弱目标探测、激光遥感、自动驾驶等领域均展现出了极大的应用潜力。为了探究如何利用该技术得到更多维度的目标信息,提出并验证了一种获取目标姿态的方法。将目标处于不同姿态时的单光子计数三维成像图(深度图)建成数据库,作为先验信息,通过求取库中图像与目标实际的单光子计数成像深度图(姿态未知)的相关系数,选取相关性最强的库中姿态作为目标实际姿态。采用单光子阵列探测器搭建实验系统,激光发散照明目标,以20°为单位构建数据库中的-60°至40°的目标深度图。结合库中数据与姿态分别处于-45°和25°时的目标深度图,利用所提方法估计目标姿态并验证其准确性。在该两种姿态下,分别做出光子计数为10,50,100时的深度图,以探究目标实际姿态与库中对应姿态的相似度和光子计数间的关系。以15°和20°为单位对目标进行多轴旋转,以探究目标进行多轴姿态变化时所提方法的可行性。改变背景噪声,在信号光子与背景光子数比值(SBR)分别为8.13,4.83,3.21,0.72的条件下对目标进行三维成像,探究背景噪声对估计成功率的影响。以木头人玩具为目标,以30°为单位建立数据库,对处于-20°和20°的木头人玩具进行成像并验证所提方法对复杂目标姿态估计的可行性。实验结果表明:所提方法可成功对目标实际姿态进行估计。目标实际姿态与库中对应姿态的相关性和光子计数成正相关,高SBR有利于准确估计目标姿态。Single-photon counting imaging technology has demonstrated potential for applications in faint target detection,laser remote sensing,and automatic driving.Therefore,in order to explore it for obtaining more dimensional target information,a target attitude acquisition method is proposed and verified.The single-photon counting three dimensional(3D)images(depth images)of the target in different attitudes are built into a database as a priori information.By calculating the correlation coefficient between the database image and the actual single-photon counting image(unknown attitude)of the target,the most relevant attitude in the database is selected as the actual attitude.A single-photon array detector is then used to establish the experimental system,and the target is illuminated by a diverging laser.The target depth images with attitudes between-60°to 40°in the database are constructed with 20°at increments.Combining database depth images and target depth images with attitudes of-45°and 25°,the proposed method is used to estimate the target attitude as well as verify its accuracy.Under these two postures,depth images with photon counts of 10,50,and 100 are made,respectively,to explore the relationship between the similarity and photon count between the actual posture of the target and the corresponding posture in the database.The multi-axis rotation of the target was conducted at 15°and 20°to explore the proposed method's feasibility when the multi-axis change of target attitude occurs.To explore the influence of background noise on estimation success rate,the background noise was changed and 3D imaging was conducted for the target under the condition that the signal photon counts to background photon counts ratio(SBR)is 8.13,4.83,3.21,and 0.72,respectively.Using wood man as the target,the database is established with 30°as the unit,and the wood man at-20°and 20°is imaged while the proposed method's feasibility for complex target attitude estimation is verified.Experimental results show that the proposed

关 键 词：三维成像 激光雷达 姿态估计 时间相关单光子计数 

分 类 号：TN29[电子电信—物理电子学]