Laser & Optoelectronics Progress, Volume. 56, Issue 2, 021502(2019)
Long Time Target Tracking Based on Kernel Correlation Filtering
Figures & Tables(6)
Fig. 2. Precision curves and success rate curves of four trackers. (a) Precision; (b) success rate
Fig. 3. Precision curves and success rate curves of four contrast algorithms under occlusion condition. (a) Precision; (b) success rate
Table 1. Video sequence
View tableTable 1. Video sequence
Video Frame Target size /(pixel×pixel) Influence factor Car scale 252 26×42 Scale variation, occlusion Crossing 120 50×17 Illumination variation, background clutter David3 252 131×35 Occlusion, background clutter, rotation Lemming 1336 103×61 Occlusion, Scale variation, rotation Jogging 307 101×25 Occlusion, background clutter Subway 175 51×19 Occlusion, rotation Walking 412 79×24 Scale variation, deformation Warlking2 500 115×31 Scale variation, occlusion, low resolution
Table 2. Performance comparison of different tracking algorithms
View tableTable 2. Performance comparison of different tracking algorithms
Algorithm CLE VOR Frame rate /(frame·s-1) CSK 80.78 0.389 186.2 KCF 35.32 0.499 233.2 CN 62.78 0.401 141.8 Ours 30.02 0.589 46.08
Tools
Get Citation
Copy Citation Text
Jianfeng Yang, Jianpeng Zhang. Long Time Target Tracking Based on Kernel Correlation Filtering[J]. Laser & Optoelectronics Progress, 2019, 56(2): 021502
Paper Information
Category: Machine Vision
Received: Jun. 1, 2018
Accepted: Aug. 8, 2018
Published Online: Aug. 1, 2019
The Author Email: Yang Jianfeng (jfyang@mail.lzjtu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20