A novel fast target tracking based on video image

摘要

Accurate motion prediction of moving target in video images is mainly studied in this paper. Real-time tracking system of high frame rate cannot accept big cost of calculation time. Therefore it needs to improve the tracking speed of target image processing in the premise of ensuring the accuracy. A method named image correlation with Kalman prediction is proposed in this paper. The improved template matching algorithm is given, and the target motion model is established which is used to predict trajectories. The advantage of this method is to use the matching results to modify the predicting data, which can achieve the fast of the algorithm and the accuracy of the tracking at the same time. Through various noise simulations, the validity of this algorithm is confirmed by the experiment given in this paper. The results show that this novel algorithm has advantages of good performance of real time and anti-interference.