Visual tracking of numerous targets via multi-Bernoulli filtering of image data

被引：149

作者：

Hoseinnezhad, Reza ^{[1
]}

Ba-Ngu Vo ^{[2
]}

Ba-Tuong Vo ^{[2
]}

Suter, David ^{[3
]}

机构：

[1] RMIT Univ, Melbourne, Vic, Australia

[2] Univ Western Australia, Nedlands, WA 6009, Australia

[3] Univ Adelaide, Adelaide, SA 5005, Australia

来源：

PATTERN RECOGNITION | 2012年 / 45卷 / 10期

基金：

澳大利亚研究理事会;

关键词：

Random finite sets; Multi-target tracking; Visual tracking; Track-before-detect; PARTICLE FILTER; JOINT DETECTION; ALGORITHM; OBJECTS;

D O I：

10.1016/j.patcog.2012.04.004

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

This paper presents a novel Bayesian method to track multiple targets in an image sequence without explicit detection. Our method is formulated based on finite set representation of the multi-target state and the recently developed multi-Bernoulli filter. Experimental results on sport player and cell tracking studies show that our method can automatically track numerous targets, and it outperforms the state-of-the-art in terms of false positive (false alarm) and false negative (missing) rates as detection error measures, and in terms of label switching rate and lost tracks ratio as tracking error measures. (C) 2012 Elsevier Ltd. All rights reserved.

引用

页码：3625 / 3635

页数：11

共 39 条

[1] Multiple target tracking with lazy background subtraction and connected components analysis [J].