Using The Side of The Current Block to Search The Optimal Block for Motion Estimation
碩士 === 義守大學 === 資訊管理學系碩士班 === 96 === The newest compression protocol of videos, H.264/AVC, had been recently proposed. There are many improvements on compression technologies in which the motion estimation has seven models, 16×16、16×8、8×16、8×8、8×4、4×8 and 4×4, that can be selected. When macroblocks...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
2008
|
Online Access: | http://ndltd.ncl.edu.tw/handle/41027979514790712986 |
id |
ndltd-TW-096ISU05396032 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-096ISU053960322015-10-13T14:52:51Z http://ndltd.ncl.edu.tw/handle/41027979514790712986 Using The Side of The Current Block to Search The Optimal Block for Motion Estimation 以目前區塊邊緣搜尋最佳預估區塊之運動估測 Chun-Hao Yeh 葉鈞豪 碩士 義守大學 資訊管理學系碩士班 96 The newest compression protocol of videos, H.264/AVC, had been recently proposed. There are many improvements on compression technologies in which the motion estimation has seven models, 16×16、16×8、8×16、8×8、8×4、4×8 and 4×4, that can be selected. When macroblocks are divided smaller and smaller, the estimations of the small blocks will be more and more accurate. However, the bit number of motion vector (MV) and reference frame will be increased in the video stream. For solving the problem, in the paper we will propose a motion estimation (ME) algorithm to estimate the position of the estimation of the current block. The algorithm utilizes the side region of the current block instead of the current block to do motion estimation in the search window. It is clear that the small blocks have higher correlation than the big blocks in the temporal and spatial images. Appropriately using the characteristic can reduce the bit number of MV and reference frame to be sent. In order to keep the video quality, the full search is also used in the ME proposed. As we mentioned above, when the proposed method obtain equal or better estimation of the current block than the full search, and only one flag bit is transmitted to locate the position of the best estimation of the current block and need not transmit the bits of reference frame and motion vector because the ME we proposed can be synchronously operated in encoder and decoder. Results of the experiments show that in different video rates, 30 and 15, the PSNR of the proposed method is better than the full search about 0.249 and 0.187 dB. Jyi-Chang Tsai 蔡吉昌 2008 學位論文 ; thesis 51 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 義守大學 === 資訊管理學系碩士班 === 96 === The newest compression protocol of videos, H.264/AVC, had been recently proposed. There are many improvements on compression technologies in which the motion estimation has seven models, 16×16、16×8、8×16、8×8、8×4、4×8 and 4×4, that can be selected. When macroblocks are divided smaller and smaller, the estimations of the small blocks will be more and more accurate. However, the bit number of motion vector (MV) and reference frame will be increased in the video stream. For solving the problem, in the paper we will propose a motion estimation (ME) algorithm to estimate the position of the estimation of the current block. The algorithm utilizes the side region of the current block instead of the current block to do motion estimation in the search window. It is clear that the small blocks have higher correlation than the big blocks in the temporal and spatial images. Appropriately using the characteristic can reduce the bit number of MV and reference frame to be sent. In order to keep the video quality, the full search is also used in the ME proposed. As we mentioned above, when the proposed method obtain equal or better estimation of the current block than the full search, and only one flag bit is transmitted to locate the position of the best estimation of the current block and need not transmit the bits of reference frame and motion vector because the ME we proposed can be synchronously operated in encoder and decoder. Results of the experiments show that in different video rates, 30 and 15, the PSNR of the proposed method is better than the full search about 0.249 and 0.187 dB.
|
author2 |
Jyi-Chang Tsai |
author_facet |
Jyi-Chang Tsai Chun-Hao Yeh 葉鈞豪 |
author |
Chun-Hao Yeh 葉鈞豪 |
spellingShingle |
Chun-Hao Yeh 葉鈞豪 Using The Side of The Current Block to Search The Optimal Block for Motion Estimation |
author_sort |
Chun-Hao Yeh |
title |
Using The Side of The Current Block to Search The Optimal Block for Motion Estimation |
title_short |
Using The Side of The Current Block to Search The Optimal Block for Motion Estimation |
title_full |
Using The Side of The Current Block to Search The Optimal Block for Motion Estimation |
title_fullStr |
Using The Side of The Current Block to Search The Optimal Block for Motion Estimation |
title_full_unstemmed |
Using The Side of The Current Block to Search The Optimal Block for Motion Estimation |
title_sort |
using the side of the current block to search the optimal block for motion estimation |
publishDate |
2008 |
url |
http://ndltd.ncl.edu.tw/handle/41027979514790712986 |
work_keys_str_mv |
AT chunhaoyeh usingthesideofthecurrentblocktosearchtheoptimalblockformotionestimation AT yèjūnháo usingthesideofthecurrentblocktosearchtheoptimalblockformotionestimation AT chunhaoyeh yǐmùqiánqūkuàibiānyuánsōuxúnzuìjiāyùgūqūkuàizhīyùndònggūcè AT yèjūnháo yǐmùqiánqūkuàibiānyuánsōuxúnzuìjiāyùgūqūkuàizhīyùndònggūcè |
_version_ |
1717759339614175232 |