Optical Flow improvement towards real time and natural rigid human motion estimation

The implementation of Differential Optical Flow Algorithms in detecting human motion still faces great challenges. To date, there are no general approaches that are suitable especially when dealing with dynamic robust environment. As such, in this paper we propose a method that combines the simple partial derivative adopted from Lukas Kanade, the regularization technique by Horn Schunck and the choice of iteration using number of level implemention in Brox Warping. Results show good segmentation, constant flow of control and improve execution time. In sum, it can be concluded that the proposed method is effective and can be used in a robust environment.