A mode and search range decision scheme for fast motion estimation in H.264

  • Authors:

  • Chien-Li Lin;Jin-Jang Leou

  • Affiliations:

  • Department of Computer Science and Information Engineering, National Chung Cheng University, Chiayi, Taiwan, Republic of China;Department of Computer Science and Information Engineering, National Chung Cheng University, Chiayi, Taiwan, Republic of China

  • Venue:
  • ICCOM'05 Proceedings of the 9th WSEAS International Conference on Communications
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this study, a mode and search range decision scheme for fast motion estimation in H.264 is proposed. The proposed scheme is used to speed up the motion estimation (ME) procedure in H.264 with slight video quality degradation. For a general video sequence, most of the macroblocks (MBs) belonging to the background (foreground) in a video frame are usually the "background" (foreground) MBs in the next video frame. Thus, the minimum matching errors of MBs in a video frame may be predicted by that of the corresponding MBs in the previously-encoded video frame so that the ME operation of an MB may be performed over a reduced set of inter modes and a reduced search range. Based on the QP value and the minimum matching errors of the MBs within the previously-encoded video frame, two thresholds, Tbackground and Tforeground, are use to classify each MB into one of three classes. Different sets of inter modes will be selected for different MB classes. Additionally, if the maximum magnitude of the x and y components of the MV(s) of the variable size block(s) within the corresponding MB in the previously-encoded video frame of an MB in the current video frame, denoted as MVmm, can be used to determine the reduced search range, Rreduced, of its integer pixel ME operation. Compared with five fast motion estimation algorithms, the proposed scheme usually has the best performance (the minimum ME processing time) in most simulation cases.