Affiliation:
1. Shandong Luneng Intelligent Technology Co., Ltd.
2. Yanbian University
Abstract
Taking the wavelet decomposed approximate image as the main research object, a direction estimation method for moving object was proposed in this paper. Firstly, the approximate image for the frame of the video was obtained via wavelet decomposition; and furthermore, the motion estimation on the approximate image was achieved to obtain the motion vectors. Finally, the motion vectors were described as polar coordinate form to compute the number of motion vectors in specified angles and the information entropy of the motion directions. The experiment results show that the proposed method can remove the effect of noise and the results of direction estimation are consistent with the actual motion directions.
Publisher
Trans Tech Publications, Ltd.
Reference10 articles.
1. Shobhit Saxena, François Brémond, Monnique Thonnat, Ruihua Ma. Crowd Behavior Recognition for Video Surveillance[C]. 10th International Conference, ACIVS, France, 2008. 971-978.
2. Junyan Tong . The Study and Analysis of Crowd Motion in Surveillance Videos[D]. Nanjing University of Science and Technology, (2009).
3. Dongping Zhang, Huailiang Peng, Yu Haibin and Yafei Lu. Crowd Abnormal Behavior Detection Based on Machine Learning[J]. Information Technology Journal, 2013 12 (6) 1199-1203.
4. Youjun Xiang, Zongze Wu, Shengli Xie. Fast Motion Estimation Algorithm Based on Moving Direction Prediction[J]. Computer Engineering, 2009 35 (24) 20-22.
5. Amer I., Abul-ainine K., Badawy W. Comparisons and Analysis of Motion Estimation Search Algorithms[J]. World Congress on Computer Science and Information Engineering. Los Angeles/Anaheim, USA, 2009 601-604.