基于主动红外视觉探测的激光跟踪仪目标跟踪恢复方法

Laser tracker target tracking recovery method based on detection system with active infrared vision

  • 摘要: 为实现激光跟踪仪的自主快速跟踪恢复,提出了一种基于主动红外视觉的跟踪恢复方法。首先,对所采用的跟踪恢复原理进行了分析。其次,充分利用激光跟踪仪合作目标靶球的强反射特性,设计了主动红外视觉探测系统。该系统采用红外SLED作为主动光源,利用SLED大发散角、红外相机的视场角远大于激光跟踪仪PSD探测范围的综合优势,实现合作目标靶球的大范围主动探测。然后,对红外图像中目标的快速识别与定位算法进行了研究,提出了基于合作目标特征评分的快速识别方法。最后,构建了跟踪恢复系统实验装置,并在算法的处理速度与识别准确度、系统的跟踪恢复范围和跟踪恢复速度等方面开展了针对性验证实验。实验结果表明,算法的平均处理速度约为每秒28帧,同时95.4%的目标中心像素坐标定位偏差低于5个像素值;在合作目标靶球距离3 m的情况下,跟踪恢复系统能够在1.8 s内实现直径0.5 m空间范围内的跟踪恢复。

     

    Abstract: In order to realize the automatic fast tracking recovery of laser tracker, a tracking recovery method based on active infrared vision was proposed. Firstly, the principle of tracking recovery was analyzed. Secondly, the active infrared vision detection system was designed by making full use of the strong reflection characteristic of the cooperative target ball of laser tracker. The system adopted infrared SLED as the active light source. The divergence angle of SLED light source was larger and the view angle of infrared camera was far greater than the detection range of PSD of laser tracker. Based on the above comprehensive advantages, the system can achieve a wide range of active detection for cooperative target ball. Then, the fast target recognition and location algorithm was studied, and a fast target recognition method based on cooperative target feature score was proposed. Finally, an experimental device of the tracking recovery system was set up. The verification experiments were carried out in the aspects of processing speed and recognition accuracy of the algorithm, tracking recovery range and speed of the system. The experimental results show that the average processing speed of the algorithm is about 28 frames per second, and the positioning deviation of 95.4% target center pixel is less than 5 pixel values. When the cooperative target ball is 3 meters away from the system, the tracking recovery system can achieve tracking recovery within the space of 0.5 meters which is completed in 1.8 seconds.

     

/

返回文章
返回