Context multi-task visual object tracking via guided filter

作者：

Highlights：

•

• A multi-task sparse learning tracker is proposed.

• We integrate context information into subspace learning method.

• The guided filter is employed to alleviate the effect of illumination and background clutter.

• The decomposition model is utilized to separate the common part and outliers in the candidates.

摘要

•A multi-task sparse learning tracker is proposed.•We integrate context information into subspace learning method.•The guided filter is employed to alleviate the effect of illumination and background clutter.•The decomposition model is utilized to separate the common part and outliers in the candidates.

论文关键词：Visual object tracking,Context information,Multi-task sparse learning,Guided filter,Alternating direction method of multipliers

论文评审过程：Received 9 August 2017, Revised 18 November 2017, Accepted 19 December 2017, Available online 1 January 2018, Version of Record 10 January 2018.

论文官网地址：https://doi.org/10.1016/j.image.2017.12.008