Human Movement Detection And Tracking Algorithm Based On Video Images

Detection and tracking of the body targets in video image is a fundamental step in many vision system including intelligent surveillance and humen-machine interface.The design and implementation of algorithms for detection and tracting of body targets are given in this paper.A common solutation for detection of moving objects in video images is to use Gaussian mixture model to represent per pixel,but it suffers form solw learing at the beginning,in-additon,a promblem of this approache is balancing between speed and stabillity. This paper present a effective scheme which reconstruct the background model using frame substracti-on.we get the background model from pixel which keeping stability in the several images of frame substraction. A comparision has been made between the two methods,result show that our system learn faster as well as better segmentation.Meanshift algorithm doesn't consider the target's motion direction and speed information, and it easily fails to track the target when affected by complex environment or occlusion dist-urbence.on the one hand,this paper propse an algorithm combined Kalman filter with Meansh-ift altorithm,the position of the target can be predicted by the Kalman of the approache, it can solve defects above.on the other hand,considering the Meanshift applay to track with certain spatial aspectst,an improve method that based on adaptive features select is presented in this paper.by combining the hue space and gray gradiant space and the current feature space was selected dynamically according to the distance defined in this paper.inaddition,a practical trac-king system is required to update the appearance change of moving object,a intergal projecti-ion has been introduced into humen targets in this paper.The experiental testing demonstrated that the tracking algorithm worked better than Combining Meanshift and Kalman algorithm in complex environment and occlusion case.it also can realize the real-time humen targets track-ing while ensuring the tracking veracity.