【作者】 齐美彬； 张祺； 蒋建国； 孙洪艳； 夏娜；

【Author】 Qi Meibin, Zhang Qi, Jiang Jianguo, Sun Hongyan, Xia Na(School of computer and Information, Hefei University of Technology, Hefei 230009,China)

【机构】 合肥工业大学计算机与信息学院；

【摘要】 本文设计了一种基于TI公司TMS320DM642数字信号处理器的实时视频主动跟踪系统,提出了一种在摄像机运动情况下的目标跟踪算法——在卡尔曼递推公式中加入控制量的方法,实现了自动调整摄像机视野跟踪运动目标。算法通过帧差法检测运动目标;再用坐标映射法对二值化图像进行目标分割;如果场景中出现多个运动目标,摄像机根据所关注目标的MNI值进行目标匹配;运用卡尔曼滤波预测系统的运动状态,得到最优控制量,并以此调整云台对目标实施实时定位跟踪。更多还原

【Abstract】 In this paper, we designed a real-time active tracking system to video motion object based on TI’s TMS320DM642 DSP. We proposed a new object tracking algorithm under the condition of the camera moving, a method which added the control factors in Kalman iterative formulae. The system could automatically adjust the camera’s scope to track the motion object. This algorithm detected motion object by temporal difference and divided the motion objects from the binary image by coordinates mapping. If there were multi-objects, the camera could match the interested object by NMI feather. The Kalman filter algorithm was operated to predict the movement state of the system and get the best control factors, by which the system adjusted the cradle head to track the object.更多还原