关键词: 点目标, 序列图像, 最大模糊熵高斯聚类, 跟踪

Abstract: A new algorithm is described for real-time tracking in image sequences based on the maximum fuzzy entropy Gaussian clustering：In the case that the initial information（position，velocity） of the target are known，it adopts the maximum fuzzy entropy Gaussian clustering to fuse the measured points in tracking window，then the fused point will be inputted to the Kalman filter to predict the next state，then continue creating a tracking window at the predicted state，detect the possible points，fuse them，until all the images have been tracked.Theoretical and experimental results show that in image sequence this algorithm can keep track of real-time，while providing high tracking accuracy.

Key words: dim point target, image sequences, maximum fuzzy entropy Gaussian clustering, tracking