Fast Multiple Reference Frame Motion Estimation for H.264 Based on Qualified Frame Selection Scheme

碩士 === 國立臺北科技大學 === 電機工程系所 === 93 === Multiple reference frame motion compensation has been adopted by the emerging video coding standard H.264/AVC. It improves the inter-frame prediction by compensating blocks from more than one reference frames. However, the process is computationally involved and...

Full description

Bibliographic Details
Main Authors: Huang-Bin Chen, 陳皇斌
Other Authors: Tien-Ying Kuo
Format: Others
Language:en_US
Published: 2005
Online Access:http://ndltd.ncl.edu.tw/handle/22s367
id ndltd-TW-093TIT05442034
record_format oai_dc
spelling ndltd-TW-093TIT054420342019-05-31T03:35:54Z http://ndltd.ncl.edu.tw/handle/22s367 Fast Multiple Reference Frame Motion Estimation for H.264 Based on Qualified Frame Selection Scheme 適用於H.264之參考畫面選擇機制與快速動態估測法 Huang-Bin Chen 陳皇斌 碩士 國立臺北科技大學 電機工程系所 93 Multiple reference frame motion compensation has been adopted by the emerging video coding standard H.264/AVC. It improves the inter-frame prediction by compensating blocks from more than one reference frames. However, the process is computationally involved and should be sped up for practical realization. This work aimed to this problem and proposed a fast algorithm, called Fast Multiple Reference Frame Motion Estimation (FMRFME). The principle of our proposed FMRFME is to qualify reference frames by the preliminary results of the continuous tracking search, and only the qualified frames should be further tested in variable block size motion estimation. The experimental results show that the proposed method reduces a considerable amount of complexity of multiple reference frame motion estimation while keeping similar R-D performance as full search. Tien-Ying Kuo 郭天穎 2005 學位論文 ; thesis 28 en_US
collection NDLTD
language en_US
format Others
sources NDLTD
description 碩士 === 國立臺北科技大學 === 電機工程系所 === 93 === Multiple reference frame motion compensation has been adopted by the emerging video coding standard H.264/AVC. It improves the inter-frame prediction by compensating blocks from more than one reference frames. However, the process is computationally involved and should be sped up for practical realization. This work aimed to this problem and proposed a fast algorithm, called Fast Multiple Reference Frame Motion Estimation (FMRFME). The principle of our proposed FMRFME is to qualify reference frames by the preliminary results of the continuous tracking search, and only the qualified frames should be further tested in variable block size motion estimation. The experimental results show that the proposed method reduces a considerable amount of complexity of multiple reference frame motion estimation while keeping similar R-D performance as full search.
author2 Tien-Ying Kuo
author_facet Tien-Ying Kuo
Huang-Bin Chen
陳皇斌
author Huang-Bin Chen
陳皇斌
spellingShingle Huang-Bin Chen
陳皇斌
Fast Multiple Reference Frame Motion Estimation for H.264 Based on Qualified Frame Selection Scheme
author_sort Huang-Bin Chen
title Fast Multiple Reference Frame Motion Estimation for H.264 Based on Qualified Frame Selection Scheme
title_short Fast Multiple Reference Frame Motion Estimation for H.264 Based on Qualified Frame Selection Scheme
title_full Fast Multiple Reference Frame Motion Estimation for H.264 Based on Qualified Frame Selection Scheme
title_fullStr Fast Multiple Reference Frame Motion Estimation for H.264 Based on Qualified Frame Selection Scheme
title_full_unstemmed Fast Multiple Reference Frame Motion Estimation for H.264 Based on Qualified Frame Selection Scheme
title_sort fast multiple reference frame motion estimation for h.264 based on qualified frame selection scheme
publishDate 2005
url http://ndltd.ncl.edu.tw/handle/22s367
work_keys_str_mv AT huangbinchen fastmultiplereferenceframemotionestimationforh264basedonqualifiedframeselectionscheme
AT chénhuángbīn fastmultiplereferenceframemotionestimationforh264basedonqualifiedframeselectionscheme
AT huangbinchen shìyòngyúh264zhīcānkǎohuàmiànxuǎnzéjīzhìyǔkuàisùdòngtàigūcèfǎ
AT chénhuángbīn shìyòngyúh264zhīcānkǎohuàmiànxuǎnzéjīzhìyǔkuàisùdòngtàigūcèfǎ
_version_ 1719197150984798208