To propose a gesture model updating and results forecasting algorithm based on Mean Shift, and to solve the problem of target model changing and influenced tracking results in gesture target tracking process. Firstly, the background difference and skin color detection methods are used to detect and get gesture model, and the Mean Shift algorithm is used to track gesture and update the gesture model, and finally to use the Kalman algorithm to predict the gesture tracking results. The

more »

... results show that this algorithm reduces the influence of surrounding environment in gesture tracking process, and get better tracking result. Index Terms-Mean shift algorithm, Hand tracking, Gesture model, Result forecasting I. . His research interests mainly include digital video compression and communications, multi-view video coding, etc. J for the DWT noted that a sensible choice depends primarily on the time series at hand-the same also holds for the MODWT. It is interesting to note, however, that, if 2 J N = , we cannot set 0 J to be greater than J for the DWT (because the DWT pyramid algorithm must terminate when there is but a single scaling coefficient left), whereas in theory there is nothing to stop us from picking 0 J J > for the MODWT (because the MODWT pyramid algorithm always yields N scaling coefficients for use at the next level). Setting 0 J J > would superficially seem to offer a way of obtaining information about variations in our times series over scales greater than the entire extent of the series, which is counter-intuitive. When 2 J N = , all the elements of 0 J S  must be equal to the sample mean X for all 0 J J ≥ . Since 0 1 0 0 1 J J J