This paper describes different approaches of realtime GMM (Gaussian Mixture Method) background subtraction algorithm using video sequences for hand image segmentation. In each captured image, the segmentation takes place where pixels belonging to the hands are separated from the background based on background extraction and skin-color segmentation. A time-adaptive mixture of Gaussians is used to model the distribution of each pixel color value. For an input image, every new pixel value is checked, deciding if it matches with one of the existing Gaussians based on the distance from the mean in terms of the standard deviation. The best matching distribution parameters are updated and its weight is increased. It is assumed that the values of the background pixels have low variance and large weight. These matched pixels, considered as foreground, are compared based on skin color thresholds. The hands position and other attributes are tracked by frame. That enables us to distinguish the hand movement from the background and other objects in movement, as well as to extract the information from the movement for dynamic hand gesture recognition.