期刊文献+

基于改进LK光流的目标跟踪算法研究 被引量:5

Research on the object tracking based on improved LK algorithm
下载PDF
导出
摘要 为了可以在复杂环境下对运动目标进行有效的跟踪,因此对LK(Lucas-Kanade)算法进行了研究并改进。改进的算法采用LK图像配准算法检测目标的状态,可以满足算法对实时性的要求;采用前向后向光流法作为LK框架,使得算法具有自评估能力,不易陷入局部极小值;将压缩感知理论引入LK算法进行目标跟踪,通过对校准误差的L1范数进行最小化来计算目标的状态参数,使得算法可以应对目标外观的变化;引入卡尔曼滤波器作为运动模型,先对物体的位置进行预测,然后再进行分类和检测,能有效地应对由于模板更新而产生的漂移问题。通过大量的实验对算法进行验证,该算法与现有算法相比处理目标在跟踪过程中的外观变化以及环境中的遮挡、光照变化等问题时,仍有较好的性能。 Inorder to track moving object in complex scenes,this paper made some extended on the LK( Lucas-Kanade) algorithm.Using LK image registration algorithm searches for the object state,which can meet the realtime requirement of the algorithm.It uses the forward backward optical flow method as the LK framework,which makes the algorithm not easy to fall into local minimization. Using the compressed sensing theory in the algorithm,and it solves the state parameters of the object through the L1 norm minimization calibration error,which makes the algorithm can be a very good deal with object appearance change. The algorithm introduces Kalman filter as a motion model to predict the location of objects,so the algorithm can effectively deal with the problems of drift due to template update. Through a lot of experiments show that the algorithm has the ability to deal with the changes in appearance,occlusion,illumination changes. Compared with other algorithm,it has a better performance.
作者 张忠义
出处 《信息技术》 2015年第10期127-130,共4页 Information Technology
关键词 目标跟踪 LK光流法 压缩感知 卡尔曼滤波 object tracking LK optical flow compressed sensing Kalman filter
  • 相关文献

参考文献11

  • 1Zhang K,Zhang L,Yang M. Real-Time Compressive Tracking[ C ]// Proceedings of European Conference on Computer Vision. 2012, Part Ⅲ: 866 - 879.
  • 2Ross D,Lim J,Lin R S,et al. Incremenlal learning for robust visual tracking[ J]. International Journal of Computer Vision, 2008, 77(1): 125-141.
  • 3Mei Xue,Ling Hai-bin. Robust visual tracking and vebicle classifi- cation via sparse representation [ J ]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33 ( 11 ) :2259 - 2272.
  • 4Babenko B,Ming Hsuan Y,Belongie S. Visual tracking with online multiple instance learning[ C ]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition,2009:983 -990.
  • 5Grabnerh, Grabnerm, Bischofh. Semi-supervised on-line boosting for robust tracking[ C ]//Proceedings of European Conference on Com- puter Vision,2008:234 - 247.
  • 6Stalders, Grabnerh, VanGool. Beyond semi-supervised tracking: tracking should be as simple as detection, but not simpler than recog- nation[ C]//Proceedings of the 12th [EEE International Conleonce on Computer Vision,2009 : 1409 - 1416.
  • 7Bertsekas D. Constrained Optimization and Lagrange Multiplier Methods[ M ]. Boston : Academic Press, 1982.
  • 8Kalal Z, Mikolajczyk K, Matas J. Forward-Backward ernor : automatic detection of tracking failures [ C ]//Pruceedings of the 201h Interna- tional Conference on Pattern Recognition. Istanbul:IEEE Computer Society,2010 : 2756 - 2759.
  • 9Donoho DL. Compressed scnsing[ J ]. IEEE Transactions on hffor- mation Theory,2006,52(4) 1289 - 1306.
  • 10Ng, A, Jordan. On discriminative vs. generative elassifier: a com- parison of logistic regression and naive hayes [ C ]//Prcweedings of Advances in Neural Information Processing Systems,2002:841 - 848.

同被引文献28

引证文献5

二级引证文献6

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部