ETRI-Knowledge Sharing Plaform

ENGLISH

성과물

논문 검색
구분 SCI
연도 ~ 키워드

상세정보

학술대회 Analysis of Camera Operations in MPEG Compressed Domain Based on Generalized Hough Transform
Cited 2 time in scopus Download 1 time Share share facebook twitter linkedin kakaostory
저자
유원영, Joonwhoan Lee
발행일
200110
출처
Pacific Rim Conference on Multimedia (PCM) 2001 (LNCS 2195), pp.1102-1107
DOI
https://dx.doi.org/10.1007/3-540-45453-5_151
협약과제
01MR4500, 디지털 컨텐츠 관리기술 개발, 홍진우
초록
In this paper, we propose a simple and efficient method to estimate the camera operations using MVs(Motion Vectors), which is extracted directly from MPEG-2 video stream without complete decoding. In the method, MVs in the compressed video stream are converted into approximated OF(Optical Flow). And the approximated OF is used to estimate the camera operations including pan, tilt and zoom based on the generalized Hough transform technique. The method provided better results than the conventional least square method for video streams of basketball and soccer games including captions. And these camera operations can be used in constructing a key frame as a mosaics image or making features for content-based video retrievals.
KSP 제안 키워드
Compressed Domain, Content based, Generalized Hough transform, Key frame, Least Squares(LS), Least-square method, MPEG-2, Motion Vector(MV), Optical Flow, compressed video stream