摘要
为实现远距离对可疑目标的跟踪功能,提高红外制导的引导精度,提出了一种多抽头流水线架构的多目标跟踪方式。该方法将图像预处理、图像分割和目标检测等算法均在FPGA内部通过卷积运算实现,通过提高卷积运算速度达到实时处理的效果。试验结果表明,该方法与其他跟踪算法相比,可以自适应调整窗口大小、改变跟踪策略和自动分配窗口,提高了跟踪的多样性和可选性。该方法可以引导红外制导系统对多个目标同时进行跟踪,最终给出判据对威胁等级高的目标进行打击,具有实际应用前景。
To achieve long-range tracking of suspicious targets and improve the accuracy of infraredguidance,a multi-tap pipeline architecture multi-target tracking mode is proposed. This method through convolution of image preprocessing, image segmentation and object detection algorithms in the FPGA. By increasing the speed of the convolution achieve the real-time processing. Experimental results show that this method compared with other tracking algorithm, can adaptively adjust the window size, change tracking policy and automatic allocation window which improve the tracking of diversity and choice. The method can guide the infrared guidance system for tracking multiple targets simultaneously, Final criterion is given for the high level of threat against targets. It has practical application prospect.
出处
《电子测量技术》
2017年第3期103-107,共5页
Electronic Measurement Technology
关键词
多目标跟踪
红外制导
多抽头流水线
multi-target tracking
infrared guidance, multi-tap line
