Conventional treatment of visual tracking has been to optimize an objective function in a probabilistic framework. In this formulation, efficient algorithms employing simple prior distributions are usually insufficient to handle clutters (e.g., Kalman filter). On the other hand, distributions that are complex enough to incorporate all a priori knowledge can make the problem computationally intractable (e.g., particle filters (PF)). This paper proposes a new formulation of visual tracking where every piece of information, be it from a priori knowledge or observed data, is represented by a set in the solution space and the intersection of these sets, the feasibility set, represents all acceptable solutions. Based on this formulation, we propose an algorithm whose objective is to find a solution in the feasibility set. We show that this set theoretic tracking algorithm performs effective face tracking and is computationally more efficient than standard PF-based tracking.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Real-time tracking with multiple cues by set theoretic random search


    Contributors:
    Cheng Chang, (author) / Ansari, R. (author)


    Publication date :

    2005-01-01


    Size :

    401510 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    TRACKING OBJECTS WITH MULTIPLE CUES

    CHEN I-KUEI | European Patent Office | 2019

    Free access

    TRACKING OBJECTS WITH MULTIPLE CUES

    CHEN I-KUEI | European Patent Office | 2019

    Free access

    Tracking objects with multiple cues

    CHEN I-KUEI | European Patent Office | 2021

    Free access

    Using Multiple Cues for Hand Tracking and Model Refinement

    Lu, S. / Metaxas, D. / Samaras, D. et al. | British Library Conference Proceedings | 2003