Partitioning and multi-feature adaptive fusion multi-object visual tracking

SHI Ying; GAO Jian-po; CUI Jie; SUN Li-juan; CHEN Yu

doi:10.13873/J.1000-9787(2015)09-0128-04

Chinese

您当前的位置：

首页 >

文章列表页 >

Partitioning and multi-feature adaptive fusion multi-object visual tracking

更新时间：2025-09-11

- Partitioning and multi-feature adaptive fusion multi-object visual tracking
- Vol. 34, Issue 9, Pages: 128-131(2015)
- 作者机构：
  
  南京理工大学电光院
- 作者简介：
- 基金信息：
- DOI：10.13873/J.1000-9787(2015)09-0128-04
  CLC： TP391.41
- Published：2015
- 稿件说明：
移动端阅览
[1]施滢,高建坡,崔杰,孙丽娟,陈宇.分块多特征自适应融合的多目标视觉跟踪[J].传感器与微系统,2015,34(09):128-131.

SHI Ying, GAO Jian-po, CUI Jie, et al. Partitioning and multi-feature adaptive fusion multi-object visual tracking[J]. 2015, 34(9): 128-131.
[1]施滢,高建坡,崔杰,孙丽娟,陈宇.分块多特征自适应融合的多目标视觉跟踪[J].传感器与微系统,2015,34(09):128-131. DOI： 10.13873/J.1000-9787(2015)09-0128-04.

SHI Ying, GAO Jian-po, CUI Jie, et al. Partitioning and multi-feature adaptive fusion multi-object visual tracking[J]. 2015, 34(9): 128-131. DOI： 10.13873/J.1000-9787(2015)09-0128-04.

摘要

为提高多目标视觉跟踪算法的实时性和稳定性

提出了分块多特征融合的目标跟踪算法。该算法融合底层颜色、纹理和边缘特征信息

以降低单一目标特征算法容易受复杂环境和目标形变的影响。建立分块目标多特征融合直方图模型

引入目标和背景区分度抑制背景分量

并且结合Kalman滤波器进行预测

在发生遮挡时根据置信度最大子块位置获取遮挡目标位置

实现目标稳定可靠的跟踪。实验结果表明:该算法对每帧图像的平均处理时间为36.2 ms

达到实时性的目的

且算法鲁棒性较强。

Abstract

In order to improve real-time and stability of multiple objects visual tracking algorithm

a target tracking algorithm based on fragment and multi-feature fusion is proposed. This algorithm fuses features of bottom layer color

texture and edge

in order to reduce single target feature algorithm is easily influenced by complex environment and target deformation. Build fragment-based target histogram with multi-feature fusion

introduce distinction degree of target and back-ground to restrain background component and combined with Kalman filter to forecast in the case of occlusion

according to the most confidence fragment for keep out object location

the algorithm achieves reliable goal tracking. Experimental results show that average processing time of the algorithm for every frame is 36. 2 ms

and achieve goal of real-time

and algorithm has strong robustness.

关键词

Keywords

references

融合SIFT特征的压缩跟踪算法 [J]. 钟权,周进,崔雄文. 光电工程 . 2015(02)

分布式粒子滤波算法在目标跟踪中的应用 [J]. 谷静,史健芳. 传感器与微系统 . 2014(08)

基于颜色纹理直方图的带权分块均值漂移目标跟踪算法 [J]. 李冠彬,吴贺丰. 计算机辅助设计与图形学学报 . 2011(12)

基于粒子滤波和GVF-Snake的目标跟踪算法 [J]. 董春利,董育宁,刘杰. 仪器仪表学报 . 2009(04)

多目标跟踪的改进Camshift/卡尔曼滤波组合算法 [J]. 孙凯,刘士荣. 信息与控制 . 2009(01)

基于特征角点的目标跟踪和快速识别算法研究 [J]. 王向军,王研,李智. 光学学报 . 2007(02)

An adaptive color-based particle filter [J] . Katja Nummiaro,Esther Koller-Meier,Luc Van Gool. Image and Vision Computing . 2002 (1)

Views

111

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

UAV tracking method based on multi-feature fusion and scale adaptive

Application of Kalman filtering in MEMS gyroscope measuring turning rate of ship

Fault diagnosis of wind turbine bearing based on multi-feature fusion and XGBoost

An improved multi-channel feature visual tracking algorithm

Research on noise analysis and noise reduction method of MEMS accelerometer

Related Author

PANG Zhengya

ZHOU Wei

ZHOU Zhifeng

TU Ting

LI Wenkui

CHEN Yongbing

ZHOU Gang

ZAI Deguang

Related Institution

College of Mechanical and Automotive Engineering, Shanghai University of Engineering Science

Military Representative Office in Xiangtan District

Department of Navigation Engineering,Naval University of Engineering

Key Laboratory of Advanced Process Control For Light Industry,Ministry of Education,Jiangnan University

Sunward Intelligent Equipment Co Ltd

⁰