Title :
Video stabilization as a variational problem and numerical solution with the Viterbi method
Author_Institution :
HP Labs Eur., Bristol, UK
Abstract :
This paper presents a new method for video stabilization based on a variational formulation that produces optimally stable sequences independently of the input motion but within the hard constraints imposed by finite sensor size to the maximum allowed compensatory transformation. A practical and fast numerical solution using the Viterbi algorithm is also proposed. The results show the efficacy of the method compared to a class of previous approaches.
Keywords :
dynamic programming; image motion analysis; image sequences; maximum likelihood estimation; variational techniques; video signal processing; Viterbi algorithm; Viterbi method; finite sensor; variational formulation; video sequences; video stabilization; Cameras; Europe; Filtering theory; Finite impulse response filter; Frequency; IIR filters; Image sensors; Layout; Motion analysis; Viterbi algorithm;
Conference_Titel :
Computer Vision and Pattern Recognition, 2004. CVPR 2004. Proceedings of the 2004 IEEE Computer Society Conference on
Print_ISBN :
0-7695-2158-4
DOI :
10.1109/CVPR.2004.1315090
