DocumentCode :
2720660
Title :
Research on Feature Extraction Technology of Competition Video Shot
Author :
Wang, Qian ; Liu, Bin
Author_Institution :
Dept. of Phys. Educ., Inst. of Disaster Prevention, Beijing, China
fYear :
2012
fDate :
11-13 Aug. 2012
Firstpage :
1381
Lastpage :
1384
Abstract :
The paper introduces theory of spatial scales as well as the processes of extracting planar SIFT characters in details. Meanwhile, aiming at defects of SIFT features, the paper proposes an improved SIFT characters. Firstly, extract SIFT characters for all frame images in the shot, and then screen them through the thought of spatio-temporal interest point, the process of which can separate parts of movements and background in the shot effectively. The screened SIFT characters can demonstrate video content of sports news in an effective manner. Results of the research indicate that mark precision of marking method in the paper is higher than other approaches.
Keywords :
feature extraction; transforms; video signal processing; competition video shot; feature extraction technology; frame images; improved SIFT characters; mark precision; marking method; spatial scales; spatio-temporal interest point; video content; Data mining; Feature extraction; Hidden Markov models; Image color analysis; Lenses; Support vector machines; Training; SIFT; feature extraction; video shot;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computer Science & Service System (CSSS), 2012 International Conference on
Conference_Location :
Nanjing
Print_ISBN :
978-1-4673-0721-5
Type :
conf
DOI :
10.1109/CSSS.2012.348
Filename :
6394586
Link To Document :
بازگشت