摘要
针对传统核相关滤波目标跟踪算法跟踪精确度不足问题,论文利用卷积神经网络来提取图像深度特征,克服传统特征的不鲁棒性。其次,结合改进核相关滤波目标跟踪算法,当目标发生遮挡时,能够准确对目标进行跟踪。论文选取代表性视频对算法进行了测试,并与传统特征进行了对比。结果显示该算法精确度提高17.6,算法跟踪速率达到20.59fps。
In order to solve the problem that the tracking accuracy of the traditional target tracking algorithm is insufficient,this paper uses the convolution neural network to extract the image depth features and overcome the robustness of the traditional features. Secondly,combined with the improved target tracking algorithm of nuclear correlation filtering,when the target is blocked,the target can be tracked accurately. In this paper,representative video is selected to test the algorithm and compared with traditional features. The results show that the accuracy of the algorithm is increased by 17.6 and the tracking rate reaches 20.59 fps.
引文
[1]Naiyan Wang,Jianping Shi,Dit-YanYeung,JiayaJia.Understanding and Diagnosing Visual TrackingSystems[C]//IEEE International Conference on Computer Vision(ICCV),2015:3101-3109.
[2]Wang N,Yeung D Y.Learning a deep compact image representation for visual tracking[C]//Advances in Neural Information Processing Systems,2013:809-817.
[3]Martin Danelljan,Gustav H?ger,Fahad Khan,Michael Felsberg.Learning Spatially Regularized Correlation Filters for Visual Tracking[C]//ICCV,2015:4310-4318.
[4]David Held,Sebastian Thrun,Silvio Savarese.Learning to Track at 100 FPS with Deep Regression Networks[C]//ECCV,2016:2152-2159.
[5]R.Rifkin,G.Yeo,T.Poggio.Regularizedleast-squares classification.Nato Science Series Sub Series III[J].Computer and Systems Sciences,2003,790:131-154.
[6]Henriques J,Caseiro R,Martins P,et al.Exploiting the circulant structure of tracking-by-detectionwith kernels[C]//European Conference on ComputerVision,2012:707-715.
[7]Henriques J,Caseiro R,Martins P,et al.High-speed tracking with kernelized correlation filters[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37(3):583-596.
[8]Matthias Mueller,Neil Smith,Bernard Ghanem.Context-Aware Correlation Filter Tracking.oral[C]//Conference on Computer Vision and Pattern Recognition(CVPR),2017:1387-1395.
[9]Wu Y,lim J,Yang M H.Online Object Tracking:A Benchmark[C]//IEEE Conference on Computer Vision and Pattern Recognition,2013(6):2411-2418.
[10]张雷,王延杰,孙宏海,等.采用核相关滤波器的自适应尺度目标跟踪[J].光学精密程,2016,24(2):448-459.ZHANG Lei,WANG Yanjie,SUN Honghai,et al.Adaptive Scale Target Tracking Using Nuclear Correlation Filters[J].Optical precision process,2016,24(2):448-459.
[11]余礼扬,范春晓,明悦.改进的核相关滤波目标跟踪算法[J].计算机应用,2015,35(12):3550-3554.YU Liyang,FAN Chunxiao,MING Yue.Improved Target Tracking Algorithm of Nuclear Correlation Filtering[J].Computer application,2015,35(12):3550-3554.
[12]秦晓飞,戴顺锋,李峰.基于改进的核相关滤波器的目标跟踪算法[J].测控技术,2017,36(10):3-10.QIN Xiaofei,DAI Shunfeng,LI Feng.Target Tracking Algorithm Based On Improved Nuclear Correlation Filters[J].Measurement and control technology,2017,36(10):3-10.
[13]高珊,李秀华,张峰,等.基于多目标区域的图像检索[J].电视技术,2018,42(4):55-61.GAO Shan,LI Xiuhua,ZHANG Feng,et al.Image Retrieval Based on Multi-Target Region[J].TelevisionTechnology,2018,42(4):55-61.
[14]Jianqiang Xu,Yao Lu,Jinwu Liu.Robust tracking via weighted spatio-temporal context learning[C]//2014IEEE International Conference on Image Processing(ICIP),2014:413-416.
[15]Zhihui Zheng,Haiping Wei,Bo Tang,Bin Zhou.A fast visual tracking method via spatio-temporal context learning for unmanned rotorcrafts fixed-pointed landing[C]//2016 IEEE Chinese Guidance,Navigation and Control Conference,2006-2010.