Shot Change Detection Using Structure Tensor in Spatio-temporal Slice

碩士 === 義守大學 === 資訊工程學系 === 92 === With the approach of the digital times, we use and encounter with a great amount of multimedia information. Therefore, the automatic analysis of digital images and sounds is a key technique. Its relative application, such as MPEG-7 and XML, are widely used now. The...

Full description

Bibliographic Details
Main Authors: Yung-Shen Lin, 林永申
Other Authors: J. H. Jeng
Format: Others
Language:zh-TW
Published: 2004
Online Access:http://ndltd.ncl.edu.tw/handle/68724301134697443150
id ndltd-TW-092ISU00392011
record_format oai_dc
spelling ndltd-TW-092ISU003920112016-01-04T04:09:17Z http://ndltd.ncl.edu.tw/handle/68724301134697443150 Shot Change Detection Using Structure Tensor in Spatio-temporal Slice 使用時空切片之結構張量於分鏡偵測 Yung-Shen Lin 林永申 碩士 義守大學 資訊工程學系 92 With the approach of the digital times, we use and encounter with a great amount of multimedia information. Therefore, the automatic analysis of digital images and sounds is a key technique. Its relative application, such as MPEG-7 and XML, are widely used now. The video is not only to provide abundant entertainments but also to store useful multimedia information. The video retrieval becomes a very important topic and its core technique is shot detection. In the thesis, structure tensor in spatio-temporal slices is used to compute 2D histogram. The algorithm, which is derived from 2D histogram difference of horizontal centerline, is used to detect shot changes. Some improved algorithms are derived by using the feature of (1) partial derivatives, (2) partial derivatives & Smooth, (3) partial derivatives & enhanced, (4) all of them. The proposed method can efficiently detect shot changes of cut as well as the complicated ones such as fade-in, fade-out, dissolve and other fused changes. J. H. Jeng 鄭志宏 2004 學位論文 ; thesis 54 zh-TW
collection NDLTD
language zh-TW
format Others
sources NDLTD
description 碩士 === 義守大學 === 資訊工程學系 === 92 === With the approach of the digital times, we use and encounter with a great amount of multimedia information. Therefore, the automatic analysis of digital images and sounds is a key technique. Its relative application, such as MPEG-7 and XML, are widely used now. The video is not only to provide abundant entertainments but also to store useful multimedia information. The video retrieval becomes a very important topic and its core technique is shot detection. In the thesis, structure tensor in spatio-temporal slices is used to compute 2D histogram. The algorithm, which is derived from 2D histogram difference of horizontal centerline, is used to detect shot changes. Some improved algorithms are derived by using the feature of (1) partial derivatives, (2) partial derivatives & Smooth, (3) partial derivatives & enhanced, (4) all of them. The proposed method can efficiently detect shot changes of cut as well as the complicated ones such as fade-in, fade-out, dissolve and other fused changes.
author2 J. H. Jeng
author_facet J. H. Jeng
Yung-Shen Lin
林永申
author Yung-Shen Lin
林永申
spellingShingle Yung-Shen Lin
林永申
Shot Change Detection Using Structure Tensor in Spatio-temporal Slice
author_sort Yung-Shen Lin
title Shot Change Detection Using Structure Tensor in Spatio-temporal Slice
title_short Shot Change Detection Using Structure Tensor in Spatio-temporal Slice
title_full Shot Change Detection Using Structure Tensor in Spatio-temporal Slice
title_fullStr Shot Change Detection Using Structure Tensor in Spatio-temporal Slice
title_full_unstemmed Shot Change Detection Using Structure Tensor in Spatio-temporal Slice
title_sort shot change detection using structure tensor in spatio-temporal slice
publishDate 2004
url http://ndltd.ncl.edu.tw/handle/68724301134697443150
work_keys_str_mv AT yungshenlin shotchangedetectionusingstructuretensorinspatiotemporalslice
AT línyǒngshēn shotchangedetectionusingstructuretensorinspatiotemporalslice
AT yungshenlin shǐyòngshíkōngqièpiànzhījiégòuzhāngliàngyúfēnjìngzhēncè
AT línyǒngshēn shǐyòngshíkōngqièpiànzhījiégòuzhāngliàngyúfēnjìngzhēncè
_version_ 1718160128949092352