快速多特征金字塔的尺度目标跟踪方法

杨源 1; 库涛 2; 查宇飞 2; 张园强 2; 李寰宇 1

doi:10.7652/xjtuxb201610008

您当前的位置：

首页 >

文章列表页 >

快速多特征金字塔的尺度目标跟踪方法

更新时间：2025-07-09

- 快速多特征金字塔的尺度目标跟踪方法
- Fast Multi-Feature Pyramids for Scale-Adaptive Object Tracking
- 西安交通大学学报 2016年50卷第10期页码：49-56
- 作者机构：
  
  空军工程大学空管领航学院,西安,710051
- 作者简介：
- 基金信息：
  
  : 2016-01-18。作者简介: 杨源(1982—),男,副教授。基金项目: 国家自然科学基金资助项目(61472442);陕西省科技新星资助项目(2015kjxx-46)
- DOI：10.7652/xjtuxb201610008
  中图分类号： TP391
- 网络首发：2016-10-10，
  
  纸质出版：2016
- 稿件说明：
移动端阅览
杨源 1, 库涛 2, 查宇飞 2, 等. 快速多特征金字塔的尺度目标跟踪方法[J]. 西安交通大学学报, 2016,50(10):49-56.

Fast Multi-Feature Pyramids for Scale-Adaptive Object Tracking[J]. 2016, 50(10): 49-56.
杨源 1, 库涛 2, 查宇飞 2, 等. 快速多特征金字塔的尺度目标跟踪方法[J]. 西安交通大学学报, 2016,50(10):49-56. DOI： 10.7652/xjtuxb201610008.

Fast Multi-Feature Pyramids for Scale-Adaptive Object Tracking[J]. 2016, 50(10): 49-56. DOI： 10.7652/xjtuxb201610008.

摘要

为了克服目标尺度变化导致的跟踪失败问题

提出了一种快速多特征金字塔的尺度目标跟踪算法。该算法融合了梯度特征和颜色特征

提高了特征表征的维度

以便获得更多的目标表征信息; 同时利用多尺度特征金字塔快速地近似相邻尺度特征

得到不同尺度模板

从而平衡了由于特征维度上升带来的计算时间开销

并保证了近似的准确性; 在相关滤波框架下

综合不同尺度模板的跟踪结果

实现对目标位置和尺度的准确估计。选取4个具有尺度变化、光照变化和背景干扰的典型场景视频序列进行仿真实验

结果表明

与传统的尺度自适应核跟踪算法相比

提出的跟踪算法能够很好地适应外部环境变化

实现对尺度目标的鲁棒跟踪

同时在中心位置误差、覆盖率、精确度和成功率4个指标上优于对比算法。

Abstract

A fast scale estimation algorithm for visual tracking with feature integration is proposed to solve tracking failure from object scale changes. The gradient feature and color feature are integrated to obtain more object representation information with the increasing feature dimensions

then a fast multi-scale feature pyramid method is used to approximate the adjacent scale features to get templates in different scales

thus it is possible to balance the computation cost due to the increasing feature dimensions without accuracy loss after approximation. Combining tracking results of multi-scale templates

the object location and scale are estimated accurately by the proposed algorithm in the framework of correlation tracking 4 representative video sequences with scale changes

and illumination variations and background clusters are chosen to simulate. The experiments indicate that the proposed algorithm well adapts to environmental variations and outperforms the traditional scale-adaptive kernel correlation tracking schemes in center location error

overlap ratio

precision and success rate.

关键词

Keywords

references

查宇飞, 杨源, 王锦江, 等. 利用密度描述符对应的视觉跟踪算法 [J]. 西安交通大学学报, 2014, 48(9): 13-18.

ZHA Yufei, YANG Yuan, WANG Jinjiang, et al. A visual object tracking algorithm using dense descriptors correspondences [J]. Journal of Xi'an Jiaotong University, 2014, 48(9): 13-18.

库涛, 毕笃彦, 杨源, 等. 尺度目标的频域核回归跟踪研究 [J]. 空军工程大学学报(自然科学版), 2016, 17(2): 76-81.

KU Tao, BI Duyan, YANG Yuan, et al. Scalable object tracking based on frequency kernel regression [J]. Journal of Air Force Engineering University(Natural Science Edition), 2016, 17(2): 76-81.

毕笃彦, 库涛, 查宇飞, 等. 基于颜色属性直方图的尺度目标跟踪算法研究 [J]. 电子与信息学报, 2016, 38(5): 1099-1106.

BI Duyan, KU Tao, ZHA Yufei, et al. Scale-adaptive object tracking based on color names histogram [J]. Journal of Electronics and Information Technology, 2016, 38(5): 1099-1106.

BOLME D S, BEVERIDGE J R, DRAPER B, et al. Visual object tracking using adaptive correlation filters [C]∥IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ, USA: IEEE, 2010: 2544-2550.

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. Berlin, Germany: Springer, 2012: 702-715.

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.

DANELLJAN M, HAGER G, KHAN F, et al. Accurate scale estimation for robust visual tracking [C]∥British Machine Vision Conference. Berlin, Germany: Springer, 2014: 1-4.

LI Y, ZHU J. A scale adaptive kernel correlation filter tracker with feature integration [C]∥European Conference on Computer Vision. Berlin, Germany: Springer, 2014: 254-265.

GRAY R M. Toeplitz and circulant matrices: a review [M]. San Francisco, USA: Now Publishers Inc., 2006: 89-101.

SCHOLKOPF B, SMOLA A J. Learning with kernels: support vector machines, regularization, optimization, and beyond [M]. Boston, Massachusetts, USA: MIT Press, 2001: 57-61.

FELZENSZWALL 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.

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.

DOLLAR P, APPEL R, BELONGIE S, et al. Fast feature pyramids for object detection [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 36(8): 1532-1545.

YOO J C, HAN T H. Fast normalized cross-correlation [J]. Circuits, Systems and Signal Processing, 2009, 28(6): 819-843.

HARE S, SAFFARI A, TORR P H S. Struck: structured output tracking with kernels [C]∥IEEE International Conference on Computer Vision. Piscataway, NJ, USA: IEEE, 2011: 263-270.

BABENKO B, YANG M H, BELONGIE S. Visual tracking with online multiple instance learning [C]∥IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ, USA: IEEE, 2009: 983-990.

POGGIO T, CAUWENBERGHS G. Incremental and decremental support vector machine learning [J]. Advances in Neural Information Processing Systems, 2001, 13(5): 409-412.

BOUGUET J Y. Pyramidal implementation of the affine lucas kanade feature tracker description of the algorithm [J]. Intel Corporation Microprocessor Research Labs Tech Rep, 2000, 22(2): 363-381.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

利用密度描述符对应的视觉跟踪算法

目标跟踪中基于深度可分离卷积的剪枝方法

水蒸气掺混对氨裂解气旋流燃烧与排放特性的影响研究

融合物理信息与深度算子网络的变压器绕组热-流场快速计算方法

结构-语义-残差协同的图像压缩框架：面向隐私保护与高保真重建