摘要
针对传统核相关滤波(Kernelized Correlation Filters,KCF)跟踪算法中利用单一特征描述目标的不完善性和目标尺度不变的局限性,论文提出了一种融合快速梯度直方图特征(Fast Histogram of Oriented Gradient,FHOG)和颜色属性特征(Color Name,CN)的尺度自适应核相关滤波算法。利用主成分分析法(Principal component analysis,PCA)获得鉴别力强的颜色属性特征减少计算量,达到颜色自适应的目的,并设计尺度自适应滤波器动态调整目标尺度。尺度滤波器与平移滤波器分别单独训练、局部优化,在保证跟踪实时性的基础上提高了算法的鲁棒性。和原KCF以及其改进算法相比,该方法在外观变形、尺度变化、光照变化、背景相似干扰等情况下有很好的适应性。
To deal with the limitations of using single feature and fixed scale in the traditional Kernelized Correlation Filters,an adaptive tracking algorithm based on KCF with multi-features fusion and adaptive scale is proposed. The feature map of the target is combined color attribute and HOG features. The computation PCA is used to obtain the discriminative CN features. The adaptive scale filter is designed to dynamically adjust the target scale. The scale filter and translation filter are trained and optimized independently. This proposed tracking algorithm is robust and real-time. Moreover,it performs better than other KCF in complex factors,such as appearance variety,scale variation,illumination variation and so on.
引文
[1]Bolme D S,Beveridge J R,Draper B A,et al.Visual object tracking using adaptive correlation filters[C]//Computer Vision and Pattern Recognition(CVPR),2010IEEE Conference on.IEEE,2010:2544-2550.
[2]Henriques J F,Caseiro R,Martins P,et al.Exploiting the circulant structure of tracking-by-detection with kernels[C]//European conference on computer vision.Springer,Berlin,Heidelberg,2012:702-715.
[3]Henriques J F,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.
[4]Felzenszwalb P F,Girshick R B,McAllester D,et al.Object detection with discriminatively trained part-based models[J].IEEE transactions on pattern analysis and machine intelligence,2010,32(9):1627-1645.
[5]Danelljan M,Shahbaz Khan F,Felsberg M,et al.Adaptive color attributes for real-time visual tracking[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2014:1090-1097.
[6]Van De Weijer J,Schmid C,Verbeek J,et al.Learning color names for real-world applications[J].IEEE Transactions on Image Processing,2009,18(7):1512-1523.
[7]Li Y,Zhu J.A Scale Adaptive Kernel Correlation Filter Tracker with Feature Integration[C]//ECCV Workshops,2014(2):254-265.
[8]Danelljan M,H?ger G,Khan F,et al.Accurate scale estimation for robust visual tracking[C]//British Machine Vision Conference,Nottingham,September 1-5,2014.BM-VA Press,2014.
[9]Rifkin R,Yeo G,Poggio T.Regularized least-squares classification[J].Nato Science Series Sub Series III Computer and Systems Sciences,2003,190:131-154.
[10]Gray R M.Toeplitz and circulant matrices:A review[J].Foundations and Trends?in Communications and Information Theory,2006,2(3):155-239.
[11]Vold H,Leuridan J.High resolution order tracking at extreme slew rates,using Kalman tracking filters[R].SAETechnical Paper,1993.
[12]Okuma K,Taleghani A,De Freitas N,et al.A boosted particle filter:Multitarget detection and tracking[C]//European conference on computer vision.Springer,Berlin,Heidelberg,2004:28-39.
[13]Comaniciu D,Ramesh V,Meer P.Real-time tracking of non-rigid objects using mean shift[C]//Computer Vision and Pattern Recognition,2000.Proceedings.IEEE Conference on.IEEE,2000,2:142-149.
[14]Everingham M,Van Gool L,Williams C K I,et al.The pascal visual object classes(voc)challenge[J].International journal of computer vision,2010,88(2):303-338.
[15]Valmadre J,Bertinetto L,Henriques J F,et al End-to-end representation learning for Correlation Filter based tracking[J].arXiv preprint arXiv:1704.06036,2017.
[16]Wu Y,Lim J,Yang M H.Online object tracking:Abenchmark[C]//Computer vision and pattern recognition(CVPR),2013 IEEE Conference on.Ieee,2013,24:11-24.