A mode filtering algorithm for accelerating HEVC FME

Yiming Cao, Satoshi Goto

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

In the high efficiency video coding (HEVC), motion estimation (ME) is an essential process for inter prediction. Compared to integer ME (IME), fractional ME (FME) occupies much more encoding time in the fast search case. To reduce the computational complexity of FME, we propose a fast algorithm based on mode filtering. In the proposal, according to results of IME and neighboring largest coding units (LCU), 1 or 2 levels are selected from the original 4-level quad-tree structure, and only the selected levels will process FME. Specifically, quad-tree separation is adopted to remove the interdependence between IME and FME. Experiments on a wide range of video sequences show that the proposed algorithm can reduce up to 42.9% and an average of 33.4% total encoding time for HEVC with little loss in PSNR and rate.

Original languageEnglish
Title of host publication2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013
Pages218-223
Number of pages6
DOIs
Publication statusPublished - 2013
Event2013 IEEE 15th International Workshop on Multimedia Signal Processing, MMSP 2013 - Pula, Sardinia
Duration: 2013 Sep 302013 Oct 2

Other

Other2013 IEEE 15th International Workshop on Multimedia Signal Processing, MMSP 2013
CityPula, Sardinia
Period13/9/3013/10/2

Fingerprint

Motion estimation
Image coding
Computational complexity
Experiments

ASJC Scopus subject areas

  • Signal Processing

Cite this

Cao, Y., & Goto, S. (2013). A mode filtering algorithm for accelerating HEVC FME. In 2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013 (pp. 218-223). [6659291] https://doi.org/10.1109/MMSP.2013.6659291

A mode filtering algorithm for accelerating HEVC FME. / Cao, Yiming; Goto, Satoshi.

2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013. 2013. p. 218-223 6659291.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Cao, Y & Goto, S 2013, A mode filtering algorithm for accelerating HEVC FME. in 2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013., 6659291, pp. 218-223, 2013 IEEE 15th International Workshop on Multimedia Signal Processing, MMSP 2013, Pula, Sardinia, 13/9/30. https://doi.org/10.1109/MMSP.2013.6659291
Cao Y, Goto S. A mode filtering algorithm for accelerating HEVC FME. In 2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013. 2013. p. 218-223. 6659291 https://doi.org/10.1109/MMSP.2013.6659291
Cao, Yiming ; Goto, Satoshi. / A mode filtering algorithm for accelerating HEVC FME. 2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013. 2013. pp. 218-223
@inproceedings{0e98dfff31074aeba1caa03f905689ac,
title = "A mode filtering algorithm for accelerating HEVC FME",
abstract = "In the high efficiency video coding (HEVC), motion estimation (ME) is an essential process for inter prediction. Compared to integer ME (IME), fractional ME (FME) occupies much more encoding time in the fast search case. To reduce the computational complexity of FME, we propose a fast algorithm based on mode filtering. In the proposal, according to results of IME and neighboring largest coding units (LCU), 1 or 2 levels are selected from the original 4-level quad-tree structure, and only the selected levels will process FME. Specifically, quad-tree separation is adopted to remove the interdependence between IME and FME. Experiments on a wide range of video sequences show that the proposed algorithm can reduce up to 42.9{\%} and an average of 33.4{\%} total encoding time for HEVC with little loss in PSNR and rate.",
author = "Yiming Cao and Satoshi Goto",
year = "2013",
doi = "10.1109/MMSP.2013.6659291",
language = "English",
isbn = "9781479901258",
pages = "218--223",
booktitle = "2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013",

}

TY - GEN

T1 - A mode filtering algorithm for accelerating HEVC FME

AU - Cao, Yiming

AU - Goto, Satoshi

PY - 2013

Y1 - 2013

N2 - In the high efficiency video coding (HEVC), motion estimation (ME) is an essential process for inter prediction. Compared to integer ME (IME), fractional ME (FME) occupies much more encoding time in the fast search case. To reduce the computational complexity of FME, we propose a fast algorithm based on mode filtering. In the proposal, according to results of IME and neighboring largest coding units (LCU), 1 or 2 levels are selected from the original 4-level quad-tree structure, and only the selected levels will process FME. Specifically, quad-tree separation is adopted to remove the interdependence between IME and FME. Experiments on a wide range of video sequences show that the proposed algorithm can reduce up to 42.9% and an average of 33.4% total encoding time for HEVC with little loss in PSNR and rate.

AB - In the high efficiency video coding (HEVC), motion estimation (ME) is an essential process for inter prediction. Compared to integer ME (IME), fractional ME (FME) occupies much more encoding time in the fast search case. To reduce the computational complexity of FME, we propose a fast algorithm based on mode filtering. In the proposal, according to results of IME and neighboring largest coding units (LCU), 1 or 2 levels are selected from the original 4-level quad-tree structure, and only the selected levels will process FME. Specifically, quad-tree separation is adopted to remove the interdependence between IME and FME. Experiments on a wide range of video sequences show that the proposed algorithm can reduce up to 42.9% and an average of 33.4% total encoding time for HEVC with little loss in PSNR and rate.

UR - http://www.scopus.com/inward/record.url?scp=84892500378&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84892500378&partnerID=8YFLogxK

U2 - 10.1109/MMSP.2013.6659291

DO - 10.1109/MMSP.2013.6659291

M3 - Conference contribution

SN - 9781479901258

SP - 218

EP - 223

BT - 2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013

ER -