Title :
Fast global motion estimation for global motion compensation coding
Author :
He, Yuwen ; Feng, Bo ; Yang, Shiqiang ; Zhong, Yuzhuo
Author_Institution :
Dept. of Comput. Sci. & Technol., Tsinghua Univ., Beijing, China
Abstract :
Global motion estimation is applied widely in computer vision. However it is difficult to get the balance of the speed and precision of global motion estimation. The fast algorithm of global motion estimation used in the global motion compensation coding is investigated in the paper. The method of global motion estimation proposed in this paper is based on dense estimation. Gauss-Newton and Levenberg-Marquadet optimizing methods are utilized. Good feature selection and robustness analysis of global motion estimation is introduced in order to accelerate the speed and guarantee the accuracy of estimation. Comparative experiments are performed to validate the performance of the proposed algorithm. The effectiveness and improvements can be observed from die comparisons
Keywords :
computer vision; feature extraction; image coding; motion compensation; motion estimation; Gauss-Newton optimizing method; Levenberg-Marquadet optimizing method; computer vision; dense estimation; die comparisons; feature selection; global motion compensation coding; global motion estimation; robustness analysis; Acceleration; Cameras; Image coding; Least squares methods; MPEG 4 Standard; Motion compensation; Motion estimation; Newton method; Recursive estimation; Robustness;
Conference_Titel :
Circuits and Systems, 2001. ISCAS 2001. The 2001 IEEE International Symposium on
Conference_Location :
Sydney, NSW
Print_ISBN :
0-7803-6685-9
DOI :
10.1109/ISCAS.2001.921050