PROJECT TITLE :
Robust Local Optical Flow for Feature Tracking
This paper is motivated by the problem of native motion estimation via sturdy regression with linear models. So as to extend the robustness of the motion estimates, we have a tendency to propose a novel sturdy local optical flow approach based mostly on a changed Hampel estimator. We tend to show the deficiencies of the smallest amount squares estimator used by the quality Kanade–Lucas–Tomasi (KLT) tracker when the assumptions created by Lucas–Kanade are violated. We propose a strategy to adapt the window sizes to address the generalized aperture downside. Finally, we evaluate our method on the Middlebury and MIT dataset and show that the algorithm provides glorious feature tracking performance with solely slightly increased computational complexity compared to KLT. To facilitate more development, the presented algorithm can be downloaded from http://www.nue.tu-berlin.de/menue/forschung/projekte/rlof.
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here