| Abstract: |
| Foreground-background separation (FBS) has been widely used in many applications, such as video surveillance and robotics. Due to the presence of the static background, a motion video can be decomposed into a low-rank background and a sparse foreground. Many regularization techniques that preserve low-rankness of matrices can therefore be imposed on the background. In the meanwhile, geometry-based regularizations, such as graph regularizations, can be imposed on the foreground. In this talk, I will present a dual-graph regularized FBS method based on weighted nuclear norm regularization and discuss its fast algorithm based on the matrix CUR decomposition. Numerical experiments on realistic human motion data sets are used to demonstrate the proposed effectiveness and robustness in separating moving objects from background, and the potential in robotic applications. |
|