Moving targets detection based on improved single Gaussian background model
Enzeng Dong, Bo Han, Xiao Yi Yu, Shengzhi Du · 2018
In order to improve the detection effect of the single Gaussian background model (SGM), an improved SGM algorithm is proposed to detect moving targets in this paper, which has made three improvements on the base of SGM. Firstly, the algorithm adopts adaptive background learning rate instead of using a fixed learning rate. Secondly, we use a new update strategy of Gaussian background model, which makes the Gaussian background model has good convergence and stability. Finally, the moving targets are detected according to the principle of Gaussian distribution and the image morphology filtering. Experimental results show that the background model of improved SGM can adapt well to background changes, and the target detection integrity is higher than that of the traditional SGM.