计算机科学 ›› 2015, Vol. 42 ›› Issue (10): 287-291.

• 图形图像与模式识别 • 上一篇    下一篇

基于滑动扫描框的高速物体的图像实时跟踪算法

郑远力,胡志坤   

  1. 中南大学物理与电子学院 长沙410012,中南大学物理与电子学院 长沙410012
  • 出版日期:2018-11-14 发布日期:2018-11-14
  • 基金资助:
    本文受湖南省自然科学基金委员会与株洲市政府自然科学联合基金(13JJ9038),湖南省科技计划(2013GK3005)资助

Real-time Tracking Algorithm for Fast Target Based on Dynamical Scanning Boxes

ZHENG Yuan-li and HU Zhi-kun   

  • Online:2018-11-14 Published:2018-11-14

摘要: TLD(Tracking-Learning-Detection)算法是近期广受关注的单目标长期跟踪算法。该算法由跟踪器、检测器、学习器协同工作,解决了目前大部分跟踪算法在目标丢失后不能重新识别目标的问题。但是由于检测器的计算量很大,该算法的实时性较差。针对这个问题,提出了一种动态生成检测扫描框的方法。输入的图片先采用跟踪器的前后向金字塔光流法加以计算,估计出目标的大概位置。然后在该位置区域生成滑动扫描框来检测。该方法可以有效缩小检测区域,减少检测器的计算量。将改进后的算法与原始算法以及Camshift、CT(Compress Tracking)算法进行了比较实验。结果表明,对于实时摄像头监控,改进的算法比原始算法具备更快的跟踪速度和更高的跟踪准确率。对于固定的图像序列,改进的算法的精度和速度都超过Camshift、CT算法。

关键词: TLD,实时性,动态滑动扫描框

Abstract: TLD algorithm is a long-term tracking algorithm for single target.And it has drawn wide attention recently.It can recognize target even the target that has been lost.However,its real-time performance is not good because of a large number of scanning boxes.We proposed a method which can generate scanning boxes dynamically.This method can reduce the calculation time efficiently and thus make TLD suit real-time situation.Experiment were conducted to compare the performance of the improved algorithm,original algorithm,Camshift and CT(Compress Tracking) algorithms.The experiment results show that when they are applied to real-time camera,the improved algorithm has faster tracking speed and higher accuracy.When they are applied on picture sequences,the speed and accuracy of the improved algorithm are better than other algorithms.

Key words: TLD,Real-time,Dynamical scanning boxes

[1] Kalal Z,Mikolajczyk K,Matas J.Tracking-learning-detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2012,34(7):1409-1422
[2] Kalal Z,Matas J,Mikolajczyk K.Pn learning:Bootstrapping binary classifiers by structural constraints[C]∥2010 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).IEEE,2010:49-56
[3] Fernando W S P,Udawatta L,Pathirana P.Identification ofmoving obstacles with pyramidal Lucas Kanade optical flow and kmeans clustering[C]∥Third International Conference on Information and Automation for Sustainability,2007(ICIAFS 2007).IEEE,2007:111-117
[4] Kalal Z,Mikolajczyk K,Matas J.Forward-backward error:Automatic detection of tracking failures[C]∥2010 20th International Conference on Pattern Recognition (ICPR).IEEE,2010:2756-2759
[5] 江博.基于Kalman的TLD目标跟踪算法研究[D].西安:西安科技大学,2013 Jiang Bo.The research of TLD target tracking algorithm based on Kalman[D].Xi’an:Xi’an University of Science and Techno-logy,2013
[6] 高帆,吴国平,刑晨,等.TLD目标跟踪算法研究[J].电视技术,2013,7(11):70-74,202 Gao fan,Wu Guo-ping,Xing Chen,et al.TLD target tracking algorithm[J].Video Engineering,2013,7(11):70-74,202
[7] Zhang K,Zhang L,Yang M H.Real-time compressive tracking[M]∥Computer Vision-ECCV 2012.Springer Berlin Heidelberg,2012:864-877
[8] Bradski G,Kaehler A.Learning OpenCV:Computer vision with the OpenCVlibrary[M].O’Reilly Media,Inc.,2008
[9] 孙凯,刘士荣.多目标跟踪的改进Camshift/卡尔曼滤波组合算法[J].信息与控制,2009,38(1):9-14 Sun Kai,Liu Shi-rong.The optimism of multi-target tracking algorithm the combination of Camshift and Kalman[J].Information and Control,2009,38(1):9-14
[10] Babenko B,Yang M H,Belongie S.Robust object tracking with online multiple instance learning[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(8):1619-1632
[11] Grabner H,Grabner M,Bischof H.Real-Time Tracking via On-line Boosting[J].BMVC,2006,1(5):6
[12] Grabner H,Leistner C,Bischof H.Semi-supervised on-line boosting for robust tracking[M]∥Computer Vision-ECCV 2008.Springer Berlin Heidelberg,2008:234-247
[13] Adam A,Rivlin E,Shimshoni I.Robust fragments-based trac-king using the integral histogram[C]∥2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.IEEE,2006:798-805
[14] Mei X,Ling H.Robust visual tracking and vehicle classification via sparse representation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(11):2259-2272
[15] Hare S,Saffari A,Torr P H S.Struck:Structured output tracking with kernels[C]∥2011 IEEE International Conference on Computer Vision (ICCV).IEEE,2011:263-270
[16] Gall J,Lempitsky V.Class-specific hough forests for object detection[M]∥Decision Forests for Computer Vision and Medical Image Analysis.Springer London,2013:143-157
[17] Clark D E,Bell J.Multi-target state estimation and track continuity for the particle PHD filter[J].IEEE Transactions on Aerospace and Electronic Systems,2007,43(4):1441-1453
[18] Zhou Xiao-long,Li Y F,He Bing-wei,et al.GM-PHD-BasedMulti-Target Visual Tracking Using Entropy Distribution and Game Theory[J].IEEE Transactions on Industrial Informatics,2014,10(2):1064-1076
[19] Zhou Xiao-long,Li Y F,He Bing-wei.Game-Theoretical Occlusion Handling for Multi-Target Visual Tracking[J].Pattern Recognition,2013,46(10):2670-2684

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!