Moving Object Tracking Method Based on Adaptive On-line Clustering and Prediction-based Cross-correlation

Making the use of the characteristics of accuracy using normalized cross-correlation image matching, this paper introduces normalized cross-correlation into the video processing, and proposes a moving object tracking method based on prediction-based Cross-correlation. First, we get the background of the video using adaptive on-line clustering method, and then get the foreground object of the video by subtracting the background. At last, through matching the object based on motion trajectory prediction and normalized Cross-correlation method, we track the moving objects and update the foreground tracking model while tracking the object. The experimental results show that our method is not only improved in real-time performance, but also can track the moving objects accurately, although there is noisy and background disturbance or the size of the object varies a lot.