Abstract—In this paper, a new method of Gaussian Mixture
Model algorithm is proposed due to the inspiration of
knowledge-based on computer vision and model recognition
which are based on computational verb theory. This algorithm
takes the binary image profiles and contour shapes to fulfill the
foreground extraction from dynamic videos. Experiments show
that, with respect to the performance in the dynamic videos, our
algorithm is better than the algorithms used widely in other
experiments. It is more accurate and can easily follow the track
of the moving objects in the videos.
Index Terms—Foreground extraction, GMM, connectivity
analysis, computational verb theory.
The authors are with the Electronic Engineering, school of Information
Science and Technology, Xiamen University, Fujian, 361001, China (e-mail:
Lbbsmile@126.com).
[PDF]
Cite: Li Bin and Yang Tao, "Foreground Extraction from Dynamic Videos Based on
Computational Verb Theory," International Journal of Modeling and Optimization vol. 4, no. 5, pp. 362-365, 2014.
