视频中快速运动目标的自适应模型跟踪算法

Zongda Liu, Liquan Dong*, Yuejin Zhao, Lingqin Kong, Ming Liu

*此作品的通讯作者

科研成果: 期刊稿件文章同行评审

15 引用 (Scopus)

摘要

To address the problem of target loss facing existing video tracking algorithms due to high mobility of targets or rapid deformation of asymmetric rigid targets, this paper proposes a video tracking algorithm based on the correlation filtering adaptive model and the redetection mechanism for average peak-to-correlation energy (APCE). The adaptive model tracking algorithm can adjust the model in real time according to the clarity of the target area, thereby effectively ensuring the accuracy of the target tracking model. Experimental results show that integrating the adaptive model tracking algorithm into the discriminative scale space tracking (DSST) model can enhance the tracking effect of the model on highly mobile or rapidly deforming objects. While guaranteeing tracking speed, the integration also raises the average accuracy of the original DSST model by 18.3 percentage points and the success rate by 15.2 percentage points. In addition, combining the adaptive tracking algorithm with the APCE redetection mechanism can ensure the stability of the algorithm.

投稿的翻译标题Adaptive Model Tracking Algorithm for Fast-Moving Targets in Video
源语言繁体中文
文章编号1815001
期刊Guangxue Xuebao/Acta Optica Sinica
41
18
DOI
出版状态已出版 - 25 9月 2021

关键词

  • Adaptive model
  • Correlation filter
  • Machine vision
  • Model update
  • Redetection mechanism
  • Target tracking

指纹

探究 '视频中快速运动目标的自适应模型跟踪算法' 的科研主题。它们共同构成独一无二的指纹。

引用此