A Study of Temporal Gradient-Based Motion Estimation for High Frame-Rate Video

Yukihiro BANDOH  Kazuya HAYASE  Seishi TAKAMURA  Kazuto KAMIKURA  Yoshiyuki YASHIMA 

Publication
D - Abstracts of IEICE TRANSACTIONS on Information and Systems (Japanese Edition)  Vol.J90-D  No.7  pp.1641-1643
Publication Date: 2007/07/01
Online ISSN: 1881-0225
Print ISSN: 1880-4535
Type of Manuscript: Special Section LETTER (Special Section on Picture Coding and Image Media)
Category: 
Keyword: 
high frame-rate sequence,  temporal gradient-based algorithm,  motion estimation, 

Full Text(in Japanese): PDF(100.7KB)

Summary: 
Motion estimation (ME) is useful in encoding video signal efficiently. At high frame-rate sequence, frame-interval shortens. High accuracy of ME is needed in the case of using the previous frame as reference frame. In order to estimate the motion, reference frame is interpolated. However, prediction performance is saturated at certain interpolation accuracy. Meanwhile, by using frames other than previous frame as reference frame, prediction error can be reduced even if ME with integer pixel accuracy. This ME does not need frame interpolation. In this case, we should restrict the candidates of reference frames, from the standpoint of computational complexity. In this paper, we describe temporal gradient-based algorithm of ME for high frame-rate sequence.