复杂环境下基于改进DeepSORT的行人实时稳定跟踪方法  

Stable and real-time pedestrian tracking method based on improved DeepSORT under complex background

作　　者：张丽娟 张紫薇[2] 姜雨彤 李东明 胡梦达 刘英雪[2] ZHANG Li-juan;ZHANG Zi-wei;JIANG Yu-tong;LI Dong-ming;HU Meng-da;LIU Ying-xue(School of Internet of Things Engineering,Wuxi University,Wuxi 214105,China;College of Computer Science and Engineering,Changchun University of Technology,Changchun 130012,China;China North Vehicle Research Institute,Beijing 100072,China;School of Information Technology,Jilin Agricultural University,Changchun 130118,China)

机构地区：[1]无锡学院物联网工程学院,江苏无锡214105 [2]长春工业大学计算机科学与工程学院,吉林长春130012 [3]中国北方车辆研究所,北京100072 [4]吉林农业大学信息技术学院,吉林长春130118

出　　处：《液晶与显示》2023年第8期1128-1138,共11页Chinese Journal of Liquid Crystals and Displays

基　　金：吉林省生态环境厅科研项目(吉环科字第2021-07号);吉林省科技发展计划重点研发项目(No.20210204050YY);无锡学院引进人才科研启动专项经费(No.2023r004,2023r006)。

摘　　要：实时多目标跟踪算法取得了理想的跟踪性能,但大多数现有算法的跟踪速度较慢,且随着背景复杂度的增加,跟踪精度也随之降低。针对此问题,本文提出了基于在线数据关联的行人实时跟踪算法。首先,设计了核相关滤波和卡尔曼滤波双轨道预测机制,配合DeepSORT中的级联匹配形成了预测-跟踪-校准体系,使数据关联更加可靠。此外,在目标检测部分引入了注意力机制,通过强化目标对象的位置信息增强特征表示能力,从而提升跟踪的精度。该模型在MOT16数据集上的MOTA达到了66.5%、IDF1达到了64.2%、IDSW达到了641。与DeepSORT算法对比,MOTA和IDF1分别提升了13%和13.2%,IDSW下降了410。本文算法有助于解决行人实时跟踪时出现的目标误检、漏检等问题,在跟踪中对严重遮挡情况仍保持了较高的跟踪精度,在复杂环境下可以实现行人实时稳定跟踪。The real-time multiple object tracking algorithms have achieved ideal tracking performance,but the tracking speed is slow,and tracking accuracy is also decreased with the increase of background complexity in most of the recent algorithms.In terms of these issues,a real-time multiple pedestrian tracking algorithm is proposed based on online data association.First of all,the dual prediction mechanism of kernelized correlation filter and Kalman filter is designed.This mechanism forms a prediction tracking calibration system with the cascade matching in DeepSORT,which makes the data correlation more reliable.In addition,the attention mechanism is introduced in the object detection part of tracking to enhance feature representation ability by strengthening position information of the object,so as to improve racking accuracy.The experiment is carried out on MOT16 dataset,the MOTA is up to 66.5%,IDF1 is up to 64.2,IDSW is 641.Compared with DeepSORT algorithm,MOTA and IDF1 increase 13%and 13.2%respectively,and IDSW decreases 410.Experimental results show that the proposed algorithm is helpful to solve the problem of object false detection,missing detection and other problems in multiple pedestrian real-time tracking.It still maintains high tracking accuracy for severe occlusion in tracking,which can achieve real-time and stable multiple pedestrian tracking in complex background.

关 键 词：多目标跟踪 实时跟踪 YOLOv5 核相关滤波算法 DeepSORT 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]