一种改进的稀疏编码视频跟踪算法*

南京大学学报(自然科学版) ›› 2013, Vol. 49 ›› Issue (2): 176–182.

一种改进的稀疏编码视频跟踪算法*

郑剑锋，张继，王洪元**

出版日期:2015-11-02 发布日期:2015-11-02
作者简介: (常州大学信息科学与工程学院，常州，213160
基金资助:
国家自然科学基金(61070121, 60973094，江苏省自然科学基金(BK2009538)，江苏省产学研前瞻性联合研究项
日(BY2009117)

A fast tracker based on sparse coding

Zheng Jian- Feng，Zhang Ji,Wang Honk-Yuan

Online:2015-11-02 Published:2015-11-02
About author: (School of Information Science and Engineering,Changzhou University,Changzhou，213164 , China)

摘要/Abstract

摘要： 最近几年，信号的稀疏表示在图像处理、人脸识别、纹理分类等领域得到了广泛的应用.在粒
子滤波框架下，视频跟踪问题被看作是使用若干个目标模板来稀疏化线性表示候选区域的过程，并使用
“小模板”来处理目标物在视频场景中出现的各种复杂变化，这种算法过程简单，但效率很低.提出一种
改进方法，使用下采样方式降低稀疏编码的复杂度，并设计了性能良好的稀疏系数向量融合方法.实验
表明，该算法在对跟踪精度几乎没有影响的前提下，大大提升了算法的效率.

Abstract: Sparse representation has been widely used in many fields，including image processing, face recognition,
texture classification and so on. With sparse representation algorithm, Xuc et al, proposed the L_¹一tracker, and they
considered the problem of video tracking as the procedure of sparsely linear represent of candidate regions with ob-
ject templates.Trivial templates arc used to deal with complex changes of objects in video scenes.The algorithm is
very straightforward but the efficiency is very low,because of the large size of over-complete dictionary, In this pa-
per, under the framework of particle filtering tracking and L_¹一tracker, we propose an advantaged algorithm to improve
the performance of original L_¹一tracker, in which，we introduce dowrrsampling to reduce the complexity of sparse rep-
resentation,and design a fusion way to together with the well-chosen sparse coefficient vectors. Our new tracking
framework is demonstrated and the experimental results on video sequences Sylv and Dudek show that,our algorithm
is much faster than Xue’s algorithm,mcanwhile without loss of tracking accuracy.

郑剑锋，张继，王洪元**
. 一种改进的稀疏编码视频跟踪算法*[J]. 南京大学学报(自然科学版), 2013, 49(2): 176–182.

Zheng Jian- Feng，Zhang Ji,Wang Honk-Yuan. A fast tracker based on sparse coding[J]. Journal of Nanjing University(Natural Sciences), 2013, 49(2): 176–182.

参考文献

[1]Mei X, Ling H. Robust visual tracking using L1 minimization. Metaxas D. IEEE 12^th International
Conference on Computer Vision, Kyoto, Japan IEEE,2009:1~8.
[2]Mei X,Ling H,Wu Y,et al. Minimum error bound- ed efficient 11 tracker with occlusion detection. Boult
T.IEEE Conference on Computer Vision and Pattern Recognition, Colorado Springs, USA, IEEE, 2011: 1257一1264.
[3]Mei X，Ling H. Robust visual tracking and vehi-cle classification via sparse representation. IEEE
Transactions on Pattern Analysis and Machine Intelligence,2011，33(11):1一15.
[4]Candes E J，Wakin M Y. An introduction to com- pressive sampling, IEEE Transactions on Signal Pro-cessing
Magazine, 2008 , 25 ( 2) ; 21一30.
[5]Candes E J，Romberg J，TaoT.Robust uncertain-ty principles; Exact signal reconstruction from
highly incomplete frequency information, IEEE Transactions on information Theory, 2006，52 (2):489一509.
[6]Yang M, Zhang L，Yang J，et al. Robust sparse coding for face recognition, IEEE Conference on
Computer Vision and Pattern Recognition, Colo- rado Springs，USA，IEEE,201 1:625一632.
[7]Yang A,Sastry S S,Gancsh A,et al. Fast 11-min- imitation algorithms and an application in robust
face recognition; A review.Technical Report. U- niversity of California at Berkeley, 2010，UCB/ EECS-2010-13.
[8]Cevher V, Sankaranarayanan A, Duarte M F, et al. Compressive sensing for background subtraction.
Ponce J. IEEE 10th European Conference on Computer Vision,Marseille, France, Springer, 2008 ;1一14.
[9]Yilmaz A，Javed O，Shah M. Object tracking: A survey. ACM Computing Surveys，2006，38(4): 1~45.
[10]Zhang J , Wang H Y. A novel tracking method based on incremental semi-supervised discriminant analy-
sis. Journal of Nanjing University(Natural Sci- ences),2012,48(4):397-404.(张继，王洪元.一
种基于增量半监督判别分析的跟踪方法.南京大学学报(自然科学),2012,48(4):397-404).
[11]Ge J Q , Li B, Chen Q M. A region-based vehicle tracking algorithm under occlusion. Journal of
Nanjing University(Natural Sciences)，2007，43 (1);66-72.葛嘉琦，李勃，陈启美.基于图像
运动区域的车辆遮挡跟踪算法.南京大学学报(自然科学),2007,43(1);66-72).
[12]Ross D A，Lim J，Lin R S. incremental learning for robust visual tracking, International Journal of
Computer Vision.2008，77(1):125一141.
[13]Kalal Z, Mikolajczyk K，Matas J.Tracking-learn ing-detection, IEEE Transactions on Pattern A
nalysis and Machine intelligence，2012，34(7) 1409一1422.
[14]Yan C,Wang Y Q,Zhang Z Y. Eye detection and tracking based on AdaBoost and Kalman algo-
rithms. Journal of Nanjing University(Natural Sciences),2012,46(6):681-687.(严超，王元
庆，张兆扬.基于AdaBoost和Kalman算法的人眼检测与跟踪.南京大学学报(自然科学),2010, 46 (6):681一687.
[15]Rosasco L，Mosci S. Santoro M，et al. Iterative projection methods for structured sparsity regu-
larization.Technical Report. MIT，MIT-CSAII: TR-2009-050. 2009.
[16]Yang M,Zhang L，Yang J，et al. Metafacc learn ing for sparse representation based face recogni
tion. IEEE 17^th international Conference on Image Processing. Hong Kong, China, IEEE，2010:1601~1604.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed