J Shanghai Jiaotong Univ Sci ›› 2024, Vol. 29 ›› Issue (3): 400-413.doi: 10.1007/s12204-022-2484-8

• • 上一篇    下一篇

基于滤波器预测的抗遮挡目标跟踪算法

陈坤1,2,赵旭1,董春玉1,邸子超1,陈宗枝1   

  1. (上海交通大学 自动化系,上海200240;2. 上海航天控制技术研究所,上海201109)
  • 接受日期:2021-09-11 出版日期:2024-05-28 发布日期:2024-05-28

Anti-Occlusion Object Tracking Algorithm Based on Filter Prediction

CHEN Kun1,2(陈坤),ZHAO Xu1∗(赵旭),DONG Chunyu1(董春玉), DI Zichao1(邸子超),CHEN Zongzhi1(陈宗枝)   

  1. (1. Department of Automation, Shanghai Jiao Tong University, Shanghai 200240, China; 2. Shanghai Aerospace Control Technology Institute, Shanghai 201109, China)
  • Accepted:2021-09-11 Online:2024-05-28 Published:2024-05-28

摘要: 视觉目标跟踪是计算机视觉中长期受到关注的重要问题。在长期跟踪中,有效处理遮挡,特别是严重遮挡的能力,是评价目标跟踪算法性能的重要方面,对提高目标跟踪算法的鲁棒性具有重要意义。然而,大多数目标跟踪算法缺乏专门处理遮挡的机制。在遮挡情况下,由于缺少目标信息,需要根据运动轨迹预测目标位置。卡尔曼滤波和粒子滤波可以有效地根据历史运动信息预测目标运动状态。一种名为概率判别模型预测(PrDiMP)的单目标跟踪方法,基于复杂场景和遮挡中的空间注意力机制。为了提高PrDiMP的性能,引入了卡尔曼滤波、粒子滤波和线性滤波。首先,针对遮挡情况,分别引入卡尔曼滤波和粒子滤波来预测物体位置,从而替代原始跟踪算法的检测结果,并停止目标模型的递归。其次,针对复杂场景中相似物体的检测跳变问题,增加了一个线性滤波窗口。在GOT-10k、UAV123和LaSOT三个数据集上的评估结果,以及几个视频上的可视化结果表明:我们的算法在遮挡情况下提高了跟踪性能,并有效抑制了检测跳变。

关键词: 单目标跟踪, 遮挡, 卡尔曼滤波, 粒子滤波, 线性滤波, 空间注意力机制

Abstract: Visual object tracking is an important issue that has received long-term attention in computer vision.The ability to effectively handle occlusion, especially severe occlusion, is an important aspect of evaluating theperformance of object tracking algorithms in long-term tracking, and is of great significance to improving therobustness of object tracking algorithms. However, most object tracking algorithms lack a processing mechanism specifically for occlusion. In the case of occlusion, due to the lack of target information, it is necessary to predict the target position based on the motion trajectory. Kalman filtering and particle filtering can effectively predict the target motion state based on the historical motion information. A single object tracking method, called probabilistic discriminative model prediction (PrDiMP), is based on the spatial attention mechanism in complex scenes and occlusions. In order to improve the performance of PrDiMP, Kalman filtering, particle filtering and linear filtering are introduced. First, for the occlusion situation, Kalman filtering and particle filtering are respectively introduced to predict the object position, thereby replacing the detection result of the original tracking algorithm and stopping recursion of target model. Second, for detection-jump problem of similar objects in complex scenes, a linear filtering window is added. The evaluation results on the three datasets, including GOT-10k, UAV123 and LaSOT, and the visualization results on several videos, show that our algorithms have improved tracking performance under occlusion and the detection-jump is effectively suppressed.

中图分类号: