Abstract:

Mixture of Gaussians (MoG) is one of the most common background modeling methods, but its accuracy comes at the expense of time, and the effect of noise treatment is general when the background is modeled in RGB color space. To improve MoG, we propose an adaptive MoG background modeling method based on YCbCr. First of all, the modeling color space is converted from RGB into YCbCr. Secondly, adaptive selection strategy is used to determine the number of gaussian components of the MoG. Finally, we order the gaussian components according to the value of sorting key words to determine the background model. The proposed modeling method is applied to moving objects detection experiments, and the experimental results show that the proposed approach is more accurate and less time-consuming in detecting moving objects compared to MoG background modeling.

Key words: background modeling;mixture of Gaussians;YCbCr color space;adaptive selection strategy