一种基于均值偏移的自动运动分割方法

计算机科学 ›› 2013, Vol. 40 ›› Issue (8): 273-276.

一种基于均值偏移的自动运动分割方法

蒋鹏,秦娜,周艳,唐鹏,金炜东

西南交通大学电气工程学院成都610031;西南交通大学电气工程学院成都610031;西南交通大学电气工程学院成都610031;西南交通大学电气工程学院成都610031;西南交通大学电气工程学院成都610031

出版日期:2018-11-16 发布日期:2018-11-16
基金资助:
本文受中央高校基本科研业务费专项资金资助

Automatic Motion Segmentation of Sparse Feature Points with Mean Shift

JIANG Peng,QIN Na,ZHOU Yan,TANG Peng and JIN Wei-dong

Online:2018-11-16 Published:2018-11-16

摘要/Abstract

摘要： 运动分割是计算机视觉领域研究的重要内容。提出一种基于均值偏移的自动运动分割算法。该方法首先用特征点匹配关系获得特征点的运动轨迹,并以轨迹的运动向量作为特征,再用均值偏移算法对轨迹的运动向量进行聚类。均值偏移缩小相似的运动向量之间的差别,同时扩大不同运动的运动向量之间的差距。为了自动获得运动分类数,还提出了一种基于非参数核密度的自动分类方法,该方法通过估计运动向量的密度分布,用核密度图自动确定运动分类数。实验结果表明,该算法分割精度高、鲁棒性好,能够自动确定运动分类数。

关键词: 均值偏移,运动分割,核密度

Abstract: We proposed an automatic motion segmentation operating on sparse feature points.Feature points are detected and tracked throughout an image sequence,and feature points are grouped using a mean shift algorithm．The motion segmentation is driven by the density of the motion vector in feature space．The kernel density estimation is performed on the mean-shifted motion vector and the number of motion present is estimated by the number of peaks in the kernel density curve.Experimental results on a number of challenging image sequences demonstrate the effectiveness and robustness of the technique.

Key words: Mean shift,Motion segmentation,Kernel density estimation

蒋鹏,秦娜,周艳,唐鹏,金炜东. 一种基于均值偏移的自动运动分割方法[J]. 计算机科学, 2013, 40(8): 273-276. https://doi.org/

JIANG Peng,QIN Na,ZHOU Yan,TANG Peng and JIN Wei-dong. Automatic Motion Segmentation of Sparse Feature Points with Mean Shift[J]. Computer Science, 2013, 40(8): 273-276. https://doi.org/

参考文献

[1] Sajid G．Motion segmentation in videos from time of flight ca-meras[C]∥201219th International Conference on Systems,Signals and Image Processing(IWSSIP)．Vienna,Austria,April 2012
[2] Gao Ji-zhou,Yang Rui-gang,Gong Ming-lun．Video Stereoli-zation:Combining Motion Analysis with User Interaction[J]．IEEE Transactions on Visualization and Computer Graphics,2012,7(18):1079-1088
[3] Jang J,Lee H．Enhanced motion estimation algorithm with prefiltering in video compression[J]．Optical Engineering,2012,1(3):37002-37009
[4] Karavasilis V,Blekas K,Nikou C．Motion Segmentation byModel-Based Clustering of Incomplete Trajectories[J]．Lecture Notes in Computer Science,2011,6912:146-161
[5] Wang Xiao-gang,Tieu K．Learning Semantic Scene Models by Trajectory Analysis[C]∥Proceedings of European Conference on Computer Vision(ECCV)．2006
[6] Antonini G,Thiran J P．Counting Pedestrians in Video Se-quences Using Trajectory Clustering[J]．IEEE Transactions on Circuits and Systems for Video Technology,2006,16(8):1008-1020
[7] Bashir F I,Khokhar A A．Real-Time Motion Trajectory-BasedIndexing and Retrieval of Video Sequences[J]．IEEE Transactions on Multimedia,2007,9(1):58-65
[8] Leiva L A,Vidal E．Revisiting the k-means algorithm for fast trajectory segmentation[C]∥ACM SIGGRAPH 2011．New York,NY,USA,2011
[9] Sundberg P,Brox T.Occlusion boundary detection and figure ground assignment from optical flow[C]∥2011IEEE Confe-rence on Computer Vision and Pattern Recognition(CVPR)．Providence,RI,2011
[10] Harris C G,Stephens M J．A combined corrter and edge detector [A]∥Proceedings Fourth Alvey Vision Conference[C]．Manchester,UK,1988:147-l51
[11] Bradski G R,Pisarevsky V．Application in calibration,stereo,segmentation,tracking,gesture,face and object recognition[C]∥IEEE Conference on computer vision and pattern recognition．SC,USA,2000:796-797
[12] Pundlik S J．Real-Time Motion Segmentation of Sparse Feature Points at Any Speed[J]．IEEE Transactions on Systems,Man,and Cybernetics,Part B:Cybernetics,2008,8(3):731-742
[13] Comaniciu D,Meer P．Mean shift:a robust approach toward feature space analysis[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2002:24(5):603-619

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed