Improved Moving Object Detection Algorithm Based on Gaussian Mixture Model

An improved method for detecting moving objects based on Gaussian mixture model for a video monitoring system is proposed in detection research.A Gaussian mixture model is established for the background based on color information.To keep adaptive characteristics and approximation to the real signal,a weighted method is provided based on the traditional learning rate in the model updating step,which gives different weights to the mean value and the variance value correspondingly.A median filter and object spatial connectivity are used to reduce noise.Experimental results indicate that compared with the old method,the proposed method can detect the moving objects more effectively,which is a robust method.