Target Tracking ECO Method Based on Response Value Judgment

被引：0
|
作者：
Chen X. [1 ]
Wang J. [1 ]
Sun Y. [1 ]
机构：
[1] School of Mechatronic Engineering, Beijing Institute of Technology, Beijing
来源：
Beijing Ligong Daxue Xuebao/Transaction of Beijing Institute of Technology | 2023年 / 43卷 / 01期
关键词：
correlation filtering; efficient convolution operator; response value judgment;
D O I：
10.15918/j.tbit1001-0645.2022.024
中图分类号：
学科分类号：
摘要：
Target tracking ECO (Efficient Convolution Operator) method is more and more widely used in various tracking scenes because of its superior tracking performance, but it shows poor tracking effect in the face of complex engineering practical situations such as occlusion, motion blur, target deformation and background clutters. To solve this problem, the ECO method was improved, and a correlation filter response value judgment mechanism was added to determine the update time of the sample model according to the maximum response mean of the previous frames and the standard deviation of the response peak of the current frame. Comparing with the original ECO method based on the same experimental video sequence, the tracking effect of the improved ECO method was showed. On OTB2015 data set, the accuracy and success rate of the improved ECO method can reach up to 88.0% and 79.9%, 1.5% and 1.2% higher than the original ECO method respectively, especially in common engineering situations such as occlusion, motion blur and background clutters. It shows that this method can provide more flexible model updating strategy and stronger ability to adapt to the actual situation of complex engineering. © 2023 Beijing Institute of Technology. All rights reserved.
引用
收藏
页码：81 / 86
页数：5
相关论文
共 15 条
  • [1] LU Huchuan, LI Peixia, WANG dong, Overview of target tracking algorithms, Pattern Recognition and Artificial Intelligence, 31, 1, pp. 61-76, (2018)
  • [2] ZHANG Kaihua, LIU Qingshan, WU Yi, Et al., Robust visual tracking via convolutional networks without training, IEEE Transactions on Image Processing, 25, 4, pp. 1779-1792, (2015)
  • [3] LIU Ye, JING Xiaoyuan, NIE Jianhui, Et al., Context-aware three-dimensional mean-shift with occlusion handling for robust object tracking in RGB-D videos[J], IEEE Transactions on Multimedia, 21, 3, pp. 664-677, (2019)
  • [4] WANG Hailuo, WANG Bo, GAO Zhifeng, Et al., Sparse representation and particle filter based online object tracking algorithm, Transactions of Beijing Institute of Technology, 36, 6, pp. 635-640, (2016)
  • [5] WANG Guanglong, TIAN Jie, ZHU Wenjie, Et al., Non-Rigid and large displacement optical flow based on descriptor matching, Transactions of Beijing Institute of Technology, 40, 4, pp. 421-426, (2020)
  • [6] NAM H, HAN B., Learning multi-domain convolutional neural networks for visual tracking, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 4293-4302, (2016)
  • [7] WANG Jianzhong, ZHANG Chiyi, SUN Yong, Target tracking method based on fusion of triple neural network and area adaptation, Transactions of Beijing Institute of Technology, 41, 2, pp. 169-176, (2021)
  • [8] BOLME D S, BEVERIDGE J R, DRAPER B A, Et al., Visual object tracking using adaptive correlation filters, International Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2544-2550, (2010)
  • [9] LIU Fuxiang, QI He, JI Ruijing, Et al., Scale adaptive anti-occlusion target tracking algorithm, Transactions of Beijing Institute of Technology, 38, 4, pp. 376-380, (2018)
  • [10] HENRIQUES, JOAO F, CASEIRO R, Et al., High-speed tracking with kernelized correlation filters, IEEE Transactions on Pattern Analysis & Machine Intelligence, 37, 3, pp. 583-596, (2014)
12