Pose estimation using visual entropy

The trend of using visual information for pose estimation of camera becomes increasingly popular and diverse. In this paper, we propose a pose estimation based on visual information entropy. The constructed cost functions are robustness by natural and suitable to be applied in pose estimation in highly agile platforms. Especially, using the visual entropy or salient features as observation causes dramatically difference in computational time. The experiments based on real data from office environment shows that the entropy-based pose estimation using mutual information has huge potential in performance.