中文摘要：

针对目标跟踪仿真系统中观瞄光点在数字仿真图像中的实时定位问题，提出一种基于单目视觉的光点实时定位方法。首先，采用卡尔曼滤波的方法对光斑在采集图像中的粗略位置进行预测；然后，在以此粗略位置为中心的光斑区域内采用基于Hessian矩阵的方法对光斑中心坐标精确提取；最后，在建立有摄像机像平面上点到数字仿真图像上对应点的定位模型基础上，由提取的光斑中心坐标经定位模型求解实现光点定位。实验结果表明：该方法对光点定位的误差为0．4pixel，且每帧图像的光点定位时间为0．32ms。该方法可实现目标跟踪仿真系统光点的高动态、高精度实时定位。

英文摘要：

Aiming at the light-spot positioning in the digital simulation image for the target observation and aiming in the target-tracking simulation systems, a method for the light-spot positioning based on monocular vision was proposed. Firstly, the rough positions of the light-spot in the camera image plane were predicted with the Kalman filter method. Then the light-spot image region with the center of the rough positions was obtained. Secondly, the precise positions of the light-spot center in the image were extracted based on the Hessian matrix in the light-spot image region. Finally, according to the positioning model calculation which was established from the point in the camera image plane to the point in the digital simulation image, the positions of the light-spot in the digital simulation image were obtained knowing its precise positions in the camera image plane. Experiment results show that the RMS error for the light-spot positioning is 0.4 pixel, and the time of processing each image is 0.32 ms. The proposed method can locate the light-spot in real time with high dynamic and high precision in the target-tracking simulation systems.