Moving Target Tracking Algorithm Research In Video Image Sequence

In recent years, with the widely application of intelligent computer, the computer providing unprecedented convenience for human life, changed the human life way. This lead to human dependence to the computer is more and more strong, in addition to the existing functions, people began to hope that the computer can understand the world, make computer have human vision, this is the computer vision. The existing target tracking technology has been widely applied to real life. For researchers how to make more accurate tracking is the key to tracking technology.In this paper, based on the common tracking technology. Research is mainly focused on the similar background and target motion state under the condition of mutations, how to make the track is more effective. Specific research contents are as follows:(1) Study the basic knowledge of image processing and moving target tracking research background, the development trends at home and abroad. Analyzes several basic tracking method, choose the meanshift algorithm for research. Study the calculation principle of the algorithm, the recursive process, how to apply in the field of target tracking. This is the based tracking algorithm for the next step work in this paper.(2) Study the tracking moving target in the static background. The camshift algorithm, called continuously adaptive mean shift. On the basis of camshift algorithm, for the actual situation in the video, using the matching parameters from modified algorithm, make the S component saturation applied to the matching process, modify the moving object matching information. The method in the moving object appeared in the similar background conditions, can separate the target from the similar color background, ensure continue to track the target. The improved matching information calculation method is also suitable for other saturation of different video. Through the verification in the experiment video, when the ball as a moving target go though the color similar background, the improved camshift can timely follow up to the target. This video is filmed by author. The third chapter in this paper provide a detailed description and experimental analysis.(3) Make the predictability of the kalman filtering combined with the camshift algorithm. Study the kalman filter recursive process, and how to make camshift algorithm calculated motion information fusion to prediction. Improve one problem of the algorithm for moving target tracking. When moving target in regular process suddenly happen mutations, such as change movement direction, according to improve kalman filter prediction step predictive value. Study the state transition matrix calculation method according to the observed value, make the algorithm increasing system sampling frequency in mutations condition, adapt to this mutation status. The fourth chapter study this content, through the contrast verify the validity and practicability of the algorithm in this paper.Aiming at similar background and target motion state under the condition of mutations situation, improve the algorithm. Through the experiment prove that the algorithm can improve the tracking accuracy.