A Background Model Based on Gaussian Statistics.
First, According to certain amount of video sequence, the paper adopts background model based on single gaussian distribution[7].The colors of every pixels are shown by Gaussian distribution model η( x,μn ,Σn ) ,where x is grey value of
n_(th) frame, μn is average value of n_(th) background image. Σn is the covariance of n_(th)
background image. When the color value & the background value of a pixel is an excessive difference, the pixel is considered as foreground pixel, otherwise as background pixel. After foreground detection, use average value and the covariance matrix of gaussian distribution to update. Introducing update rateα ,so updating the gaussian distribution parameters can be shown as (1)and(2).