期刊文献+

基于图的半监督目标跟踪 被引量:1

Graph-based semi-supervised learning for object tracking
下载PDF
导出
摘要 针对目标跟踪方法中有限的目标样本对监督分类器的影响,提出一种基于图的在线半监督跟踪方法。首先,根据前几帧的跟踪结果分别提取反映目标当前状态和长期状态的特征以及背景的特征,共同组成标记样本集,再在当前帧通过运动模型采样候选区域作为未标记样本集。然后将每个样本分割成若干部件,以所有样本对应部件构成的集合为节点构建一张图,并独立在每张图上应用基于图的半监督分类方法给出判决分数,各部件的判决结果最终被融合并给出跟踪结果。最后,用新的跟踪结果更新标记样本集。实验证明该方法对目标和背景的动态变化有很强的鲁棒性,满足实时性要求。 In order to alleviate the effect of limited labeled instances on the supervised classifier in object tracking, a new method based on the graph-based semi-supervised learning framework is proposed for object tracking. Firstly, two kinds of positive instances representing the long term and short term information about the object respectively and negative instances representing the background are extracted around the tracking re- suits of the previous few frame. Meanwhile, the candidates are sampled by a particle filter as unlabeled in- stances. Secondly, each labeled or unlabeled instance is divided into several overlapped patches with a spatial layout. The corresponding patches of all these instances are gathered together to construct a graph. The similar ity scores of the candidates are evaluated independently over each graph. Finally, these similarities from all graphs are combined to find the tracking result and update the graphs. Empirical results on challenging video se quences demonstrate the superior performance of the proposed method in robustness and accuracy to state-of- the art methods in the literature.
出处 《系统工程与电子技术》 EI CSCD 北大核心 2016年第2期450-456,共7页 Systems Engineering and Electronics
关键词 目标跟踪 半监督学习 基于部件思想 object tracking graph semi-supervised learning part-based idea
  • 引文网络
  • 相关文献

参考文献18

  • 1Yang H, Shao L, Zheng F, et al. Recent advances and trends in visual tracking : a review [ J ]. Neurocomputing, 2011,74 ( 18 ) : 3823 - 3831.
  • 2Ross D, Lim J, Lin R S, et al. Incremental learning for robust visual traeking[J:. International Journal of Computer Vision, 2008,77(1) = 125-141.
  • 3Babenko B, Yang M H, Belongie S. Visual tracking with online multiple instance learning[C]//Proe, of the IEEE Conference on Computer Vision and Pattern Recognition, 2009= 983 -990.
  • 4Hare S, Saffari A, Torr P H S. Struck: structured output tracking with kernels[C]//Proc, of the IEEE International Conference on Computer Vision, 2011 = 263 - 270.
  • 5Zhong W, Lu H, Yang M H. Robust object tracking via sparsity- based collaborative model [ C] // Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2012 = 1838 - 184,5.
  • 6Belkin M, Niyogi P, Sindhwani V. Manifold regularization= a geometric framework for learning from labeled and unlabeled ex- amples[J]. The Journal of Machine Learning Research, 2006, 7(3)= 2399-2434.
  • 7Gao Q, Liu J, Cui K, et al. Stable locality sensitive discriminant analysis for image recognition[J]. Neural Networks, 2014, 54c (6): 49-56.
  • 8Kotropoulos Y P C, Arce G R. Ll-graph based music structure analysis[C]// Proc. of the International Society Jbr Music In- formation Retrieval Conference, 2011 =495 - 500.
  • 9Cheng B, Yang J, Yan S, et al. Learning with-graph for image analysis[J]. IEEE Trans. on Image Processing, 2010, 19(4): 858 - 866.
  • 10Gu Y, Feng K. Ll-graph semisupervised learning for hyperspectral image classificationEC:// Proc. of the IEEE Geoscience and Re mote Sensing Symposium, 2012: 1401 - 1404.

同被引文献11

引证文献1

二级引证文献12

;
使用帮助 返回顶部