| 基于数据和群体运动仿真模型迭代优化的多目标跟踪算法 |
 点此下载全文 |
| 引用本文：赵花蕊.基于数据和群体运动仿真模型迭代优化的多目标跟踪算法[J].计算技术与自动化,2025,(4):44-51 |
| 摘要点击次数: 10 |
| 全文下载次数: 2 |
|
|
| 中文摘要:在线多目标跟踪通过实时分析视频序列中的每一帧信息来估计目标轨迹。尽管深度学习方法在目标稀疏且外观特征明显的场景中表现良好，但在复杂环境下，频繁的遮挡会导致外观特征不可靠，从而影响跟踪准确性。为此，提出了一种基于数据和群体运动仿真模型多目标跟踪算法，将目标运动特征与视觉特征相结合。通过融合这两种特征，构建相似度矩阵以评估目标间的相似性。最终，利用匈牙利匹配算法对相似度矩阵中的检测目标和历史跟踪目标进行匹配，从而得到最终的跟踪结果。实验表明，该方法在在线多目标跟踪方面显著提高了性能。 |
| 中文关键词:在线多目标跟踪 群体运动仿真模型 外观特征 匈牙利匹配算法 |
| |
| Multi-object Tracking Algorithm Based on Iterative Optimization of Data and Collective Motion Simulation Model |
|
|
| Abstract:Online multi-object tracking estimates object trajectories by analyzing each frame of video sequences in real time. Although deep learning methods perform well in scenarios where objects are sparse and their appearance features are distinct, frequent occlusions in complex environments can render appearance features unreliable, thereby affecting tracking accuracy. To address this issue, this paper proposes a multi-object tracking algorithm based on data and collective motion simulation models, which combines object motion characteristics with visual features. By integrating these two types of features, a similarity matrix is constructed to evaluate the similarity between objects. Finally, the Hungarian matching algorithm is employed to match detected objects in the similarity matrix with historically tracked objects, leading to the final tracking results. Experimental results demonstrate that this method significantly improves performance in online multi-object tracking. |
| keywords:online multi-object tracking collective motion simulation model appearance features Hungarian matching algorithm |
| 查看全文 查看/发表评论 下载pdf阅读器 |
