中文摘要：

目标跟踪中运动情况的复杂性给目标跟踪任务增加了许多困难。针对此问题,提出一种将光流法和卡尔曼滤波方法相结合的目标跟踪方法。首先,通过光流法处理输入视频帧;其次,经形态学膨胀和改进的中值滤波处理,进而实现对运动目标的准确获取;最后,根据所获取的目标位置等信息,使用卡尔曼滤波方法处理后续图像序列,并对运动目标进行预测,从而实现对运动目标的有效跟踪。在两组对比实验中,所提方法的跟踪平均准确率分别达到67.83%和85.25%,实验结果表明所提方法在提高跟踪准确率的同时也使得跟踪更具主动性。

英文摘要：

The complexity of movement in object tracking brings a lot of difficulties to the object tracking. A method combining optical flow with Kalman filtering was proposed to track objects in this context. Firstly, the method of optical flow was used to deal with the video frames. Secondly, the moving target was obtained by using the both methods of morphological dilation and improved median filtering. Finally, in accordance with the obtained information of the object, Kalman filtering was used to handle the following sequence of images and to make the predictions about the object. After that, the effective tracking of moving objects was achieved by the proposed method. The average tracking accuracies of the two comparative experiments were respectively 67. 83% and 85. 25%. The experimental results show that the proposed method improves the tracking accuracy and makes the tracking more initiative.