视频序列中多运动物体检测与跟踪方法

In this paper, the image pre-processing techniques have been studied first and an improved median filter method is adopted to propose an image enhancement techniques ased on the differential evolution algorithm and to go through the experimental comparison and analysis on the images and the wavelet algorithm before and after the processing.Secondly, the paper researched on the principles of the three kinds of commonly used target detection algorithm for moving target.Through the analysis and comparison of their advantages and disadvantages, it proposed a detection method for three-frame differencing moving target based on dynamic background difference.Finally, track moving targets' trajectory obtained from the image sequences. The extended Kalman filter model is established for object tracking and to do the motion forecast. It can reduce noise disturbance and shrink the search range of characteristic and improved tracking efficiency.