ردیابی حرکت چندگانه با استفاده از فیلتر کالمن

In this paper , multiple movement tracking using kalman filter is studied which is based on combining the model of pixels of each background and a set of single hypothetical background models which is based on the length of the object , situation , speed and the distribution of color .kalman filter combines the estimation of the goal’s location based on pervious measurements and estimation of current location based on current measurements. this combination has the minimum variance. in other words , it has best accuracy. a deductive model from speed and the normal movement direction is used for determining the initial speed which is used for calculating the speed of each instance of objects in the background . this model usually is an extensive kalman filter. system , works near to the umber of frames in each second(from 24 until 40 f/s). assuming that the image resolution is adequate and the size of the objects will not go any further in comparison to object size which was used in initial calculations.here , we will consider two models of people jumping and tracking the bikers and a consequential self-regulating model is proposed that has the factors of these two models and according to receive some first frames form an image , it selects the tracking algorithm of the image with considering the models’ block diagrams. then , error between the calculated area and the true area of two models plus the accuracy of this method on these two models are calculated