Periodic Pattern Detection for Real-Time Application

Digital video stabilization approaches typically degrade their performances in presence of periodic patterns. Any kind of matching between consecutive frames is not usually able to work in presence of these kind of signals: the motion estimation engine is deceived and its performances degrade abruptly. In this paper we propose a fast fuzzy classifier able to recognize periodic and aperiodic pattern in the images that takes into account the peculiarities of digital video stabilization. Finally, the proposed classifier can be used as a filtering module in a block based video stabilization approach. Key words: Video Stabilization, periodic pattern, fuzzy classifier

• Type: Conference
• Year: 2009
• Where: WILF
• Authors: Giovanni Puglisi, Sebastiano Battiato
• Keywords: Digital Video Stabilization, Periodic Pattern, Video Stabilization, WILF 2009