Summary
International Technical Conference on Circuits/Systems, Computers and Communications
2016
Session Number:T3-6
Session:
Number:5267
Histogram-based Non-Iterative Global Motion Estimation
Kyeongbo Kong, Seungjun Shin, Woo-Jin Song ,
pp.669-672
Publication Date:2016/7/10
Online ISSN:2188-5079
DOI:10.34385/proc.61.5267
PDF download (1.3MB)
Summary:
We present a compressed-domain global motion estimation (GME) algorithm that uses a robust estimator. Conventional GME algorithms based on the M-estimator are accurate and powerful, but are slow because they use an iterative process. To speed up the GME, we propose a non-iterative algorithm that exploits an outlier rejection mask and histogram. In simulations the proposed algorithm was about five times faster than the conventional algorithm and had slightly higher estimation accuracy.