| Title |
Causal Video Stabilization Using Background-Features’ Extraction and Alpha-trimmed Mean Smoothing Filter |
| Authors |
오종근(JongGeun Oh) ; 팜민티엔(Minh-Thien Pham) ; 홍민철(Min-Cheol Hong) |
| DOI |
https://doi.org/10.5573/ieie.2019.56.3.94 |
| Keywords |
video stabilization ; alpha-trimmed mean smoothing filter ; background feature ; global motion estimation ; accumulated motion parameters |
| Abstract |
This paper proposes a method of extracting background feature points and reconstructing stabilized videos by removing distorted motion components such as hand shake. Clustering classification process is explained to remove distorted motion by using the moving distance and directionality between features of consecutive frames. Accumulated motion parameters of classified background features are defined to set a constraint which is used to remove residual distorted motion. In addition, this paper addresses a way to estimate the global motion by controlling the relaxation of the alpha-trimmed mean filter with the constraints The experimental results show that the performance of th proposed method is robust to various motion parameter distortion. |