Title :
Salient frame extraction using support vector regression and motion features
Author :
Du, Xian ; Dua, Sumeet
Author_Institution :
Dept. of Comput. Sci., Louisiana Tech Univ., Ruston, LA, USA
Abstract :
We present a new support vector regression (SVR) algorithm to extract salient frames from videos. We use optical flow to describe motion in frames and an adaptive SVR to identify the abrupt change of content in frame sequences. We show that the proposed algorithm is computationally simple and effective in detecting salient frames in video sequences.
Keywords :
feature extraction; image motion analysis; image sequences; regression analysis; support vector machines; video signal processing; SVR; frame sequence; motion feature extraction; optical flow; salient frame extraction; support vector regression; video sequence; Adaptive optics; Computer vision; Estimation; Image motion analysis; Optical distortion; Support vector machines; Videos; Salient frame selection; optical flow; support vector regression;
Conference_Titel :
Aerospace and Electronics Conference (NAECON), Proceedings of the IEEE 2010 National
Conference_Location :
Fairborn, OH
Print_ISBN :
978-1-4244-6576-7
DOI :
10.1109/NAECON.2010.5712934
