融合YOLOV5+DSST+KCF的目标检测跟踪算法在激光通信中的应用
2023,31(10):49-53
摘要：在空间激光通信中，两通信设备实现激光通信需要经过粗瞄准与精瞄准的过程，而粗瞄准过程在大部分场景下需借助光学望远镜和信标激光等辅助实现，实施不易；针对在粗瞄准过程中如何快速、准确识别出激光通信设备，并辅助完成激光通信,提出将不同目标检测算法与长时跟踪KCF算法相结合；通过对比发现，YOLOV5+DSST+KCF(YDK)算法的平均帧率为12帧，在帧率满足系统要求情况下，YDK算法的响应峰值要比原KCF算法、HAAR+DSST+KCF(HDK)算法分别提高了43.5%、31.4%，且有效解决目标尺度变化问题。
关键词：激光通信；YOLOV5算法；KCF算法；目标检测；长时追踪
Application of Target Detection and Tracking Algorithm Combining YOLOV5+DSST+KCF in Laser Communication
Abstract：In the filed of space laser communication, the realization of laser communication by two communication devices needs to go though the process of coarse and fine aiming, and the process of coarse aiming needs to be assisted by tools such as optical telescopes and beacon laser in most scenarios, which is not easy to implement. Aiming at how to quickly and accurately identify laser communication devices?in the rough aiming process, and assist in the completion of laser communication, a combination of different target detection algorithms and long-term tracking KCF algorithm is proposed. Through comparison, it is found that the average frame rate of the YOLOV5+DSST+KCF(YDK) algorithm is 12 frames per second, when the frame rate meets the system requirements, the response peak value of the YDK algorithm is 43.5% and 31.4% higher than the original KCF algorithm and HAAR+DSST+KCF(HDK) algorithm, respectively, and effectively solved the problem of target scale change.
Key words：laser communication; Yolov5 algorithm; KCF algorithm; target detect; long-term tracking
收稿日期：2022-12-20
基金项目：
