Title :
A new motion estimation algorithm based on balanced multiwavelets vector transform
Author :
Wang, Bing ; Zhao, Rongchun
Author_Institution :
Inst. of Comput., Northwestern Polytech. Univ., Xi´´an, China
fDate :
31 Aug.-4 Sept. 2004
Abstract :
A new algorithm for accurate motion estimation (ME) using omnidirectionally balancing multiwavelets approximation is proposed. In gradient-based methods, the computation of motion field depends on minimizing the energy function and smoothness constraints. The proposed approach takes advantage of the nature of multiwavelets theory, which can simultaneously possess desirable properties to model vector fields and image related functions. Each flow vector and image function will be represented by linear combinations of multiscales basis functions. Once all the corresponding quadratic and convex function scale coefficients are determined, the motion field can be obtained. Compared with two existing algorithms, the new algorithm is robust in terms of accuracy.
Keywords :
approximation theory; gradient methods; image sequences; minimisation; motion estimation; surveillance; vectors; wavelet transforms; background model; balanced multiwavelets vector transform; convex function scale coefficient; energy function minimization; gradient-based methods; motion estimation algorithm; motion field; multiscales basis functions; multiwavelets theory; omnidirectionally balancing multiwavelets approximation; optical flow; quadratic function scale coefficient; shadow removal; smoothness constraint minimization; surveillance system; Approximation algorithms; Biomedical optical imaging; Computer vision; Filters; Image motion analysis; Motion estimation; Particle beam optics; Robustness; Vectors; Wavelet transforms;
Conference_Titel :
Signal Processing, 2004. Proceedings. ICSP '04. 2004 7th International Conference on
Print_ISBN :
0-7803-8406-7
DOI :
10.1109/ICOSP.2004.1441559