Statistical methods are basically used to remove the problems encountered in the background subtraction method. This method is also used to detect changes regions according to the pixel. These methods are originally generation of background subtraction method. This method usesthe properties of pixel or group of pixels to build new background model and the statistics of the background are updated dynamically. In this method the statistics are dynamically updated for each frame [14].And then pixel constituting the foreground are compared the background pixel to detect the foreground pixel. This method is more popular due to its reliability in case of noise, low resolution or illumination changes and shadow [17]. Stauffer and Grimson [2] discuss an adaptive background mixture modeled by a mixture of Gaussians which are updated on-line through images. To determine a pixel is from background or foreground, Gaussian distribution of the mixture model for that pixel is evaluated.