基于稀疏表示多子模板的鲁棒目标跟踪算法
|
摘要
目标跟踪技术是一项富有挑战性的研究课题,在红外成像搜索、红外精确制导、智能监控、运动识别等领域有着广泛的应用。文中提出了一种基于稀疏表示多子模板的鲁棒目标跟踪算法。首先,提出一种基于自适应辨别信息的子模板选择方法,最大限度地捕捉目标的结构信息,提高模板子块的整体描述;针对直方图对光照敏感的缺点,引入了稀疏表示理论对子模板进行描述,提高模板子块的表达能力和适应能力;其次通过构造表决图的形式对目标位置进行表决和融合决策;最后设计了一种动态的子模板更新策略,来有效地应对目标外观模型的变化。在大量测试图像序列中的仿真实验表明,文中所提算法可以有效应对形变、光照变化、部分遮挡、完全遮挡以及虚假目标干扰和背景干扰,具有较高的鲁棒性。
Object tracking is a challenging research topic, which is widely used in infrared imaging search, infrared precision guidance, intelligent surveillance, motion recognition and other fields. In this paper, a robust template patches-based target tracking method with sparse representation was proposed.Firstly, the adaptive template patches selection mechanism was proposed using the discriminative information to capture the target. Then, the sparse representation was introduced to describe the patches to deal with the shortcoming of histogram′ s sensitivity to light, which expanded the application of the algorithm. Thirdly, the target location was voted and fused by constructing a voting map. Finally, a dynamic updating scheme of patches was proposed to address appearance variations. The simulation experiments of test image sequences demonstrate the robustness of the proposed tracker, which is able to deal with many challenges, such as deformation, changes of illumination, partial and total occlusions,false target jamming and background interference.
Object tracking is a challenging research topic, which is widely used in infrared imaging search, infrared precision guidance, intelligent surveillance, motion recognition and other fields. In this paper, a robust template patches-based target tracking method with sparse representation was proposed.Firstly, the adaptive template patches selection mechanism was proposed using the discriminative information to capture the target. Then, the sparse representation was introduced to describe the patches to deal with the shortcoming of histogram′ s sensitivity to light, which expanded the application of the algorithm. Thirdly, the target location was voted and fused by constructing a voting map. Finally, a dynamic updating scheme of patches was proposed to address appearance variations. The simulation experiments of test image sequences demonstrate the robustness of the proposed tracker, which is able to deal with many challenges, such as deformation, changes of illumination, partial and total occlusions,false target jamming and background interference.
引文
[1] Ma Tianyi, Zhang Hurxiang, Song Minmin, et al. Antioccluded infrared target tracking with salient feature space[J]. Infrared and Laser Engineering, 2017, 46(3):0304002.(in Chinese)
[2] He Y, Li M, Zhang J, et al. Infrared target tracking based on robust low-rank sparse learning[J]. IEEE Geoscience and Remote Sensing Letters, 2016, 13(2):232-236.
[3] Lu R, Xu W, Zheng Y, et al. Visual tracking via probabilistic hypergraph ranking[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2017, 27(4):866-879.
[4] Hager G D, Dewan M, Stewart C V. Multiple kernel tracking with SSD[C]//Proceedings of ICCV, 2004:790-797.
[5] Adam A, Rivlin E, Shimshoni I. Robust fragments-based tracking using the integral histogram[C]//Proceedings of CVPR, 2006:798-805.
[6] Bins J, Dihl L L, Jung C R. Target tracking using multiple patches and weighted vector median filters[J]. Journal of Mathematical Imaging&Vision, 2013, 45(3):293-307.
[7] Nejhum S M, Ho J, Yang M H. Online visual tracking with histograms and articulating blocks[C]//Proceedings of CVPR,2008:1-8.
[8] Yao R, Shi Q, Shen C, et al. Part-based robust tracking using online latent structured learning[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2017, 27(6):1235-1248.
[9] Mei X, Ling H. Robust visual tracking using l1 minimization[C]//Proceedings of ICCV, 2009:1436-1443.
[10] Wright J, Yang A, Ganesh A, et al. Robust face recognition via sparse representation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2009, 31(2):1-18.
[11] Wang Huigai, Li Zhengzhou, Gu Yuanshan, et al. Dim target detection method based on muti-scale adaptive sparse dictionary[J]. Infrared and Laser Engineering, 2014, 43(7):2371-2378.(in Chinese)
[12] Comaniciu D, Meer P. Mean Shift:A robust approach toward feature space analysis[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence, 2002, 24(5):603-619.
[2] He Y, Li M, Zhang J, et al. Infrared target tracking based on robust low-rank sparse learning[J]. IEEE Geoscience and Remote Sensing Letters, 2016, 13(2):232-236.
[3] Lu R, Xu W, Zheng Y, et al. Visual tracking via probabilistic hypergraph ranking[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2017, 27(4):866-879.
[4] Hager G D, Dewan M, Stewart C V. Multiple kernel tracking with SSD[C]//Proceedings of ICCV, 2004:790-797.
[5] Adam A, Rivlin E, Shimshoni I. Robust fragments-based tracking using the integral histogram[C]//Proceedings of CVPR, 2006:798-805.
[6] Bins J, Dihl L L, Jung C R. Target tracking using multiple patches and weighted vector median filters[J]. Journal of Mathematical Imaging&Vision, 2013, 45(3):293-307.
[7] Nejhum S M, Ho J, Yang M H. Online visual tracking with histograms and articulating blocks[C]//Proceedings of CVPR,2008:1-8.
[8] Yao R, Shi Q, Shen C, et al. Part-based robust tracking using online latent structured learning[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2017, 27(6):1235-1248.
[9] Mei X, Ling H. Robust visual tracking using l1 minimization[C]//Proceedings of ICCV, 2009:1436-1443.
[10] Wright J, Yang A, Ganesh A, et al. Robust face recognition via sparse representation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2009, 31(2):1-18.
[11] Wang Huigai, Li Zhengzhou, Gu Yuanshan, et al. Dim target detection method based on muti-scale adaptive sparse dictionary[J]. Infrared and Laser Engineering, 2014, 43(7):2371-2378.(in Chinese)
[12] Comaniciu D, Meer P. Mean Shift:A robust approach toward feature space analysis[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence, 2002, 24(5):603-619.