Abstract: Based on the fusion of color and gradient features,this paper implements a novel approach to real-time background subtraction.Firstly,an energy function is defined based on the fusion of color and gradient features.Secondly,the graph cuts based algorithm is employed to minimize energy function and segment the foreground.Finally,average optical flow is used to make inference about the validity of foreground regions,background models are then updated.The experimental results of different real scenes show that the proposed approach can produce real-time detection and promising results.