基于二维振镜的动态目标实时跟踪与高保真信息采集系统

Dynamic target real-time tracking and high-fidelity information acquisition system based on two-dimensional galvanometers

针对目前视觉跟踪平台研究中存在的大视野与高分辨力不可兼得以及高速运动目标跟踪难以实现等痛点问题,设计了一种基于二维振镜的新型视觉跟踪平台,可以在保证视野范围的前提下,兼顾高帧率和高分辨力特性,实现动态目标50 FPS的实时跟踪以及100倍放大后的高保真信息采集。首先,对基于二维振镜的光学系统进行几何建模,对光路变化与动态视角之间的耦合关系进行分析,完成了双相机协同系统的坐标系变换;其次,针对运动目标实时跟踪任务,采用一种基于动态模板匹配和卡尔曼滤波器的目标检测跟踪算法,以保证目标物体的快速、准确定位;最后,从动态视野范围、响应速度、跟踪目标定位精度及跟踪目标极限速度四个方面对系统性能进行评估,并选择多个典型场景对系统的应用领域进行分析。结果表明:该系统特殊的双相机协同工作模式和极快速的动态响应特性,使其很好地弥补了现有视觉跟踪平台的不足,在智能监控、智慧交通、国防军工等领域中具有广阔的应用前景和极高的研究价值。

To address the painful problems in the current research of visual tracking platform, such as the inability to have both large field of view and high resolution, and the difficulty of high-speed motion target tracking, we designed a novel visual tracking platform based on two-dimensional(2D) galvanometers. It can achieve real-time tracking of dynamic targets at 50 FPS and high-fidelity information acquisition after 100 × magnification, when we take into account the high frame rate and high resolution characteristics while ensuring the field of view. Firstly, the geometrical modeling of the optical system based on the 2D galvanometers is made, the coupling relationship between the optical path change and the dynamic angle of view is analyzed, and the coordinate transformation of the dual-camera cooperative system is completed. Secondly, for the real-time target tracking task, a target detection and tracking algorithm based on dynamic template matching and Kalman filter is adopted, which ensures the fast and accurate positioning of the target object.Finally, the performance of the system is evaluated from the four aspects of dynamic field of view, response speed, tracking target positioning accuracy and tracking target limit speed, and several typical scenarios are selected to analyze the application fields of the system. The results show that the special dual-camera cooperative working mode and extremely fast dynamic response characteristics of the system well compensate for the shortcomings that exist in a visual tracking platform, and have broad application in the fields of intelligent monitoring, intelligent transportation, and military defense. It is proved that our system has broad application prospects and high research value.