A motion estimation IP with low memory access for H.264/AVC encoder based on fully parallel hardware-oriented algorithm

Shi Hye Kim, Nam Thang Ta, Jeong Hoon Kim, Jun Rim Choi

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Scopus citations

Abstract

In this paper, we present integer and fractional motion estimation IP for H.264/AVC encoder by hardware-oriented algorithm. In integer motion engine, the reference block is used to share for consecutive current macro blocks in parallel processing which exploits data reusability and reduces off-chip bandwidth. In fractional motion engine, instead of two-step sequential refinement, half and quarter pel are processed in parallel manner in order to discard unnecessary candidate positions and double throughput. The synthesis results show that the IP core occupies 907K logic gates and achieves high throughput supporting HDTV 720p 30 fps.

Original languageEnglish
Title of host publication2010 Proceedings of 19th International Conference on Computer Communications and Networks, ICCCN 2010
DOIs
StatePublished - 2010
Event2010 19th International Conference on Computer Communications and Networks, ICCCN 2010 - Zurich, Switzerland
Duration: 2 Aug 20105 Aug 2010

Publication series

NameProceedings - International Conference on Computer Communications and Networks, ICCCN
ISSN (Print)1095-2055

Conference

Conference2010 19th International Conference on Computer Communications and Networks, ICCCN 2010
Country/TerritorySwitzerland
CityZurich
Period2/08/105/08/10

Fingerprint

Dive into the research topics of 'A motion estimation IP with low memory access for H.264/AVC encoder based on fully parallel hardware-oriented algorithm'. Together they form a unique fingerprint.

Cite this