Multi-convolutional Layer Feature Response Tracking Algorithm Combined with Re-detection Mechanism

doi:10.3778/j.issn.1673-9418.2005057

Abstract

Abstract:

Target tracking algorithms based on depth features are prone to tracking drift in fast moving targets and long-term occlusions. A multi-convolution layer feature response target tracking algorithm combined with re-detection mechanism is proposed. Firstly, the target region is detected based on the image mixture Gaussian model. Secondly, the multi-convolution layer extracts the target depth feature map according to the class activation map of the weighted gradient and trains independent correlation filters to fuse the underlying spatial features and high-level semantic features. The convolutional layer filter obtains the target response position, and then the constraint term of the re-detection mechanism is used to smooth the output response value to build a strong tracker. The parameters and weight coefficients of the model are adaptively updated to avoid parameter over-fitting in the model, achieving real-time tracking effects. The experimental results show that the algorithm has high accuracy and success rate in the complex scenarios such as severe target deformation, fast movement, and long-term occlusion.

Key words: depth feature map, strong tracker, mixed Gaussian model, detection mechanism

摘要：

针对基于深度特征的目标跟踪算法在目标快速运动、长时间遮挡容易导致跟踪漂移的问题，提出了一种结合重检测机制的多卷积层特征响应跟踪算法。首先基于图像分块的混合高斯模型检测出目标区域，其次多卷积层根据加权梯度的类激活映射提取目标深度特征图，并训练出相互独立的相关滤波器，然后融合底层空间特征和高层语义特征的卷积层滤波器得到目标响应位置，再由重检测机制约束项平滑输出响应值，从而构建出强跟踪器，最后自适应地更新模型参数和权重系数，避免模型中参数过拟合，达到实时跟踪效果。实验结果表明，该算法在目标严重形变、快速运动、长时期遮挡等复杂情景下，跟踪结果具有很高的精确度和成功率。

关键词: 深度特征图, 强跟踪器, 混合高斯模型, 检测机制

ZHANG Jing, HUANG Haomiao. Multi-convolutional Layer Feature Response Tracking Algorithm Combined with Re-detection Mechanism[J]. Journal of Frontiers of Computer Science and Technology, 2021, 15(3): 533-544.

张晶, 黄浩淼. 结合重检测机制的多卷积层特征响应跟踪算法[J]. 计算机科学与探索, 2021, 15(3): 533-544.

References

[1] LU H C, LI P X, WANG D. Visual object tracking: a survey[J]. Pattern Recognition and Artificial Intelligence, 2018, 31(1): 61-76.
卢湖川, 李佩霞, 王栋. 目标跟踪算法综述[J]. 模式识别与人工智能, 2018, 31(1): 61-76.
[2] SHAN Y G, HU W G. Review of visual object tracking al-gorithms of adaptive direction and scale[J]. Computer Eng-ineering and Applications, 2020, 56(9): 13-23.
单玉刚, 胡卫国. 尺度方向自适应视觉目标跟踪方法综述[J]. 计算机工程与应用, 2020, 56(9): 13-23.
[3] HENRIQUES J F, RUI C, MARTINS P, et al. Exploiting the circulant structure of tracking-by-detection with kernels[C]// LNCS 7575: Proceedings of the 12th European Conference on Computer Vision, Florence, Oct 7-13, 2012. Berlin, Heidel-berg: Springer, 2012: 702-715.
[4] DANELLJAN M, KHAN F S, FELSBERG M, et al. Adap-tive color attributes for real-time visual tracking[C]//Proceed-ings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbia, Jun 23-28, 2014. Washington: IEEE Computer Society, 2014: 1090-1097.
[5] 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.
[6] DANELLJAN M, HAGER G, KHAN F S, et al. Discrimina-tive scale space tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(8): 1561-1575.
[7] KALAL Z, MIKOLAJCZYK K, MATAS J. Tracking-learning-detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(7): 1409-1422.
[8] DANELLJAN M, HAGER G, KHAN F S, et al. Learning spatially regularized correlation filters for visual tracking[C]// Proceedings of the 2015 IEEE International Conference on Computer Vision, Santiago, Dec 7-13, 2015. Washington: IEEE Computer Society, 2015: 4310-4318.
[9] MA C, HUANG J B, YANG X K, et al. Hierarchical convolu-tional features for visual tracking[C]//Proceedings of the 2015 IEEE International Conference on Computer Vision, Sant-iago, Dec 7-13, 2015. Washington: IEEE Computer Society, 2015: 3074-3082.
[10] NAM H, BAEK M, HAN B. Modeling and propagating CNNs in a tree structure for visual tracking[J]. arXiv:1608.07242, 2016.
[11] DANELLJAN M, BHAT G, KHAN F S, et al. ECO: effici-ent convolution operators for tracking[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, Jul 21-26, 2017. Washington: IEEE Computer Society, 2017: 6931-6939.
[12] CHEN C, DENG Z H, GAO Y L, et al. Single target tracking algorithm based on multi-fuzzy kernel fusion[J]. Journal of Frontiers of Computer Science and Technology, 2020, 14(5): 848-860.
陈晨, 邓赵红, 高艳丽, 等. 多模糊核融合的单目标跟踪算法[J]. 计算机科学与探索, 2020, 14(5): 848-860.
[13] WANG G Y, SUN Y R, WU L, et al. Scale adaptive correla-tion tracking method for circular target based on block detec-tion[J]. Computer Engineering and Applications, 2020, 56(8): 177-184.
王国屹, 孙永荣, 吴雷, 等. 分块检测尺度自适应的圆形目标相关跟踪[J]. 计算机工程与应用, 2020, 56(8): 177-184.
[14] LI X, SHEN C H, DICK A, et al. Online metric weighted linear representations for robust visual tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(5): 931-950.
[15] ARNOT K, EMMANUEL L. Generalized linear mixed model with a penalized Gaussian mixture as a random effects dis-tribution[J]. Computational Statistics & Data Analysis, 2008, 52(7): 3441-3458.
[16] SELVARAJU R R, COGSWELL M, DAS A, et al. Grad-CAM: visual explanations from deep networks via gradient-based localization[J]. International Journal of Computer Vi-sion, 2020, 128(2): 336-359.
[17] LI K, HE F Z, CHEN X. Real-time object tracking via com-pressive feature selection[J]. Frontiers of Computer Science, 2016, 10(4): 689-701.
[18] EVERINGHAM M, VAN GOOL L, WILLIAMS C K I, et a1. The pascal visual object classes (VOC) challenge[J]. Interna-tional Journal of Computer Vision, 2010, 88(2): 303-338.
[19] WU Y X, JIA N, SUN J P. Real-time multi-scale tracking based on compressive sensing[J]. The Visual Computer, 2015, 31(4): 471-484.