Moving Target Detection Based on Improved Mixture Gauss Model

Based on mixture Gauss model to detect moving targets is easy to produce "ghosting", smear and background update problem caused by light mutation, so this paper proposes an improved mixture Gauss modeling method background update problem. This method combines the frame difference method to identify moving pixels and non-real motion pixels in the foreground image. By giving non-real pixels a larger learning rate to make them fast blending into the background, we solve the "ghost", smear and light mutations problems. The experimental results show that this algorithm can effectively detect moving targets.