Journal
IEEE ACCESS
Volume 8, Issue -, Pages 208179-208194Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/ACCESS.2020.3038792
Keywords
Target tracking; Feature extraction; Correlation; Kernel; Adaptation models; Histograms; Sun; Object tracking; histogram of oriented gradient; Kalman filter; kernel correlation filter
Categories
Funding
- Ministry of Housing and Urban-Rural Development Science and Technology Planning Project [2016-R2-060]
- National Key Research and Development Plan of China [2017YFC0804400, 2017YFC0804401]
- Jiangsu Technology Project of Housing and Urban-Rural Development [2018ZD265]
- Major Project of Natural Science Research of the Jiangsu Higher Education Institutions of China [18KJA520012]
- Xuzhou Science and Technology Plan Project [KC19197]
- School-Level Scientific Research Project of Xuzhou Institute of Technology [XKY20191070]
Ask authors/readers for more resources
The traditional kernel correlation filter (KCF) algorithm has poor tracking results in complex scenes with severe occlusion, deformation, and low resolution and cannot achieve long-term tracking. To improve the accuracy of the tracking algorithm in complex scenes, an adaptive kernel correlation filter algorithm is proposed. First, a multifeature complementary scheme is proposed that linearly weights the responses of the histogram of oriented gradient (HOG) features and color features and learns a target position estimation model to realize target position estimation. Then, an adaptive scale model for estimating the scale transformation of the target is learned by extracting the HOG features of the object. Finally, according to occlusion judgment criteria, the Kalman filter is introduced to correct the position of the tracking target. The accuracy and success rate of the proposed algorithm are verified by simulation analysis on TC-128/OTB2015 benchmarks. Extensive experimental results illustrate that the proposed tracker achieves competitive performance compared with state-of-the-art trackers. The distance precision rate and overlap success rate of the proposed algorithm on OTB2015 are 0.899 and 0.635, respectively. The proposed algorithm effectively solves the long-term object tracking problem in complex scenes. This study provides references for computer vision processing, such as image retrieval, behavior analysis, and intelligent driving.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available