基于单运动目标的PTZ检测跟踪系统设计

Design of PTZ detection and tracking system based on single moving object

针对现有PTZ系统在背景图像动态变化时,对于恶劣环境、低照度和光线不足等容忍度差,误检率高,难以同时达到准确、实时、可靠等跟踪要求的缺点,提出了一种新型PTZ检测跟踪系统.运用自适应背景差分法分割出运动目标;采用改进的OSTU算法获得动态阈值;通过帧间差的结果控制背景累计更新;使用投影法实现运动目标的精确定位.在CamShift算法基础上加入Kalman滤波进行运动预测,实现了对运动目标的PTZ实时跟踪.在VC++环境中,利用图像采集卡和快球进行高速图像采集与云台控制,以行人为运动目标进行测试.结果表明,该系统实现了运动目标的检测、定位和PTZ跟踪;从其统计的检测率和运行时间看,有很好的鲁棒性和实时性.

To solve the problem that when the background images change dynamically,the existing PTZ system has a high noise ratio,low tolerence to harsh circumstance,low light intensity,poor light,and difficulty to achieve accurate,real-time,and reliable track request,a new PTZ detection and tracking system was proposed.Moving object was divided by the auto-adapted background method.The improved OSTU algorithm was used to obtain the dynamic threshold value and background accumulation update was completed by frame difference result control.the projection was adopted in order to achieve precise positioning of moving objects,the Kalman filter was added in the CamShift algorithm foundation to complete motion prediction,and the PTZ real-time track of the movement goal was realized.In the VC＋＋environment,the image gathering card and the quick ball were adopted to gather high speed image and carry out the PTZ control,and the designed PTZ examination tracking system was tested with a pedestrian as the movement goal.The results show that the movement goal examination,the localization and the PTZ track are realized by this system.In respect of the running time and detection rate,this system show good robustness and real-time character.