3rd International ICST Conference on Performance Evaluation Methodologies and Tools

Research Article

Cross-Entropy Based Data Association for Multi Target Tracking

Download701 downloads
  • @INPROCEEDINGS{10.4108/ICST.VALUETOOLS2008.4348,
        author={Daniel Sigalov and Nahum Shimkin},
        title={Cross-Entropy Based Data Association for Multi Target Tracking},
        proceedings={3rd International ICST Conference on Performance Evaluation Methodologies and Tools},
        publisher={ICST},
        proceedings_a={VALUETOOLS},
        year={2010},
        month={5},
        keywords={data association target tracking heuristic optimization cross- entropy method Monte-Carlo methods},
        doi={10.4108/ICST.VALUETOOLS2008.4348}
    }
    
  • Daniel Sigalov
    Nahum Shimkin
    Year: 2010
    Cross-Entropy Based Data Association for Multi Target Tracking
    VALUETOOLS
    ICST
    DOI: 10.4108/ICST.VALUETOOLS2008.4348
Daniel Sigalov1,*, Nahum Shimkin1,*
  • 1: Technion – Israel Institute of Technology, Haifa, Israel, 32000
*Contact email: dansigal@tx.technion.ac.il, shimkin@ee.technion.ac.il

Abstract

Multiple-target tracking (MTT) in the presence of spuri- ous measurements poses difficult computational challenges related to the measurement-to-track data association prob- lem. Different approaches have been proposed to tackle this problem, including various approximations and heuristic op- timization tools. The Cross Entropy (CE) and the related Parametric MinxEnt (PME) methods are recent optimiza- tion heuristics that have proved useful in many combina- torial optimization problems. They are akin to evolution- ary algorithms in that a population of solutions is evolved, however the solution improvement mechanism is based on statistical methods of sampling and parameter estimation. In this work we apply the Cross-Entropy method and its recent MinxEnt variants to solve approximately the multi- scan version of the data association problem in the presence of misdetections, false alarms, and unknown number of tar- gets. We formulate the algorithms, and explore via simu- lation their efficiency and performance compared to other recently proposed algorithms.