Design of accurate detection and tracking algorithm for moving target under jitter interference

被引：0
|
作者：
Zheng P. [1 ]
Bai H. [1 ]
Li Z. [1 ]
Guo H. [1 ]
机构：
[1] College of Energy and Power Engineering, Nanjing University of Science and Technology, Nanjing
来源：
Yi Qi Yi Biao Xue Bao/Chinese Journal of Scientific Instrument | 2019年 / 40卷 / 11期
关键词：
Electronic image stabilization; Kernelized correlation filter(KCF); Lucas-Kanade optical flow method; Object detection; Object tracking;
D O I：
10.19650/j.cnki.cjsi.J1905657
中图分类号：
学科分类号：
摘要：
Aiming at the problem of poor detection accuracy for moving targets under jitter interference, a moving target detection algorithm based on optical flow method and three-frame difference method is proposed. Firstly the image stabilization algorithm based on LK (Lucas-Kanade) optical flow method is used to de-jitter the video, then the three-frame difference method is used to extract the target. Simulation results show that the PSNR (Peak Signal to Noise Ratio) value is increased by 3.6 dB after image stabilization, and the designed algorithm can accurately extract the target under jitter interference, the average processing speed on the test platform is 28 fps. At the same time, aiming at the problem that traditional KCF (Kernelized Correlation Filter) algorithm has poor tracking performance for scale-changing and partially occluded targets, an improved KCF algorithm is designed, which constructs the image pyramid of the target, then calculates the filter response on different layers of the image pyramid, finds the layer with largest response and updates the target location of next frame. Meanwhile, an occlusion detection mechanism is introduced in the algorithm, which reduces the impact of target occlusion on tracking. Simulation results indicate that the improved algorithm has stronger robustness to the scale-changing and partially occluded targets, and can achieve stable target tracking, the processing speed of the algorithm is 33 fps. Compared with the KCF algorithm, the precision of the proposed algorithm is increased by 4.2% and the success rate is increased by 11.8%. © 2019, Science Press. All right reserved.
引用
收藏
页码：90 / 98
页数：8
相关论文
共 23 条
  • [1] Sun Y.J., Yu J.Y., Wang X.M., Multiple object tracking algorithm for the complex scenario, Chinese Journal of Scientific Instrument, 40, 3, pp. 126-137, (2019)
  • [2] Wang J.Z., Lu C.H., Jiang W.W., New multi-target tracking algorithm based on conditional random field, Journal of Electronic Measurement and Instrumentation, 31, 6, pp. 909-913, (2017)
  • [3] Wan M.J., Gu G.H., Cao E.C., Et al., In-frame and inter-frame information based infrared moving small target detection under complex cloud backgrounds, Infrared Physics and Technology, 76, 5, pp. 455-467, (2016)
  • [4] Hua Y.L., Liu W.J., Moving object detection algorithm of improved gaussian mixture model, Journal of Computer Applications, 34, 2, pp. 580-584, (2014)
  • [5] Gu B., Song K.F., Qiu D.Y., Et al., Moving object detection based on improved ViBe algorithm, International Journal of Smart Home, 9, 12, pp. 225-232, (2015)
  • [6] Gao P., Sun X.G., Wang W., Moving object detection based on kirsch operator combined with optical flow, International Conference on Image Analysis and Signal Processing, pp. 620-624, (2010)
  • [7] Yin H.P., Chen B., Chai Y., Et al., Vision-based object detection and tracking: a review, Acta Automatica Sinica, 42, 10, pp. 1466-1489, (2016)
  • [8] Ge B.Y., Zuo X.Z., Hu Y.J., Review of visual object tracking technology, Journal of Image and Graphics, 23, 8, pp. 1091-1107, (2018)
  • [9] Tuzel O., Porikli F., Meer P., A bayesian approach to background modeling, Computer Society Conference on Computer Vision and Pattern Recognition, pp. 58-65, (2005)
  • [10] Wu M.J., Fu T., Liu J.P., Et al., Multi-feature target tracking based on dynamic bayesian network, Computer Engineering and Applications, 47, 30, pp. 183-187, (2011)
123