|
|
Two-stage global motion estimation based on random sampling method |
ZHENG Ya-yu, TIAN Xiang, CHEN Yao-wu |
(Institute of Advanced Digital Technology and Instrument, Zhejiang University, Hangzhou 310027, China) |
|
|
Abstract A two-stage global motion estimation method based on the random sampling strategy was proposed to estimate the global motion parameters effectively. In the first stage, each group of motion vectors was sampled randomly based on the rule of the minimum matrix condition number. Then the initial transform parameter was computed by using the least squares method. The final transform parameter was achieved by utilizing the histogram method. In the second stage, the motion vector field was compensated by the transform motion firstly. Based on a group of motion vectors sampled randomly from the compensated motion vector field, the final translational parameter was estimated by using the histogram method. Experimental results on four affine global motion models and the real video sequences show that compared with the estimation method based on the fixed sampling pattern, the proposed method can decrease the computational time markedly while achieving higher estimation accuracy.
|
Published: 26 February 2010
|
|
基于随机采样的两阶段全局运动估计
为了有效地估计全局运动参数,提出一种基于随机采样的两阶段全局运动估计方法.在第一阶段,根据矩阵条件数最小化准则随机选取每组运动矢量,利用最小二乘法计算每组初始的变换参数,使用直方图获得最终的变换参数.在第二阶段,对运动矢量场进行变换运动补偿,从已补偿运动矢量场中随机采样一组运动矢量,使用直方图估计最终的平移参数.对于4个仿射全局运动模型和真实视频序列的实验结果表明,与基于固定采样模式的估计算法相比,该方法能够有效地缩短计算时间,同时获得更好的估计精度.
|
|
[1] ZITOVA B, FLUSSER J. Image registration methods: a survey [J]. Image and Vision Computing, 2003, 21(11): 977-1000.
[2] JOZAWA H, KAMIKURA K, SAGATA A, et al. Two-stage motion compensation using adaptive global MC and local affine MC [J]. IEEE Transactions on Circuits and Systems for Video Technology, 1997, 7(1): 75-85.
[3] KIM M, CHOI J G, KIM D, et al. A VOP generation tool: automatic segmentation of moving objects in image sequences based on spatio-temporal information [J]. IEEE Transactions on Circuits and Systems for Video Technology, 1999, 9(8): 1216-1226.
[4] QIAN X M, LIU G Z. Global motion estimation from randomly selected motion vector groups and GMLM based applications [J]. Signal, Image and Video Processing, 2007, 1(3): 179-189.
[5] SU Y, SUN M T. A non-iterative motion vector based global motion estimation algorithm [C]∥ IEEE Conference on Multimedia and Expo (ICME). Taipei: IEEE, 2004: 703-796.
[6] SU Y P, SUN M T, HSU V. Global motion estimation from coarsely sampled motion vector field and the applications [J]. IEEE Transactions on Circuits and Systems for Video Technology, 2005, 15(2): 232-242.
[7] RATH G B, MAKUR A. Iterative least squares and compression based estimations for a four-parameter linear global motion model and global motion compensation [J]. IEEE Transactions on Circuits and Systems for Video Technology, 1999, 9(7): 1075-1099.
[8] DUFAUX F, MOSCHENI F. Motion estimation techniques for digital TV: a review and a new contribution [J]. Proceedings of the IEEE, 1995, 83(6): 858-876.
[9] HORN R A, JOHNSON C R. Matrix analysis [M]. Cambridge: Cambridge University Press, 1985: 291-292. |
|
Viewed |
|
|
|
Full text
|
|
|
|
|
Abstract
|
|
|
|
|
Cited |
|
|
|
|
|
Shared |
|
|
|
|
|
Discussed |
|
|
|
|