Compressive Online Robust Principal Component Analysis with Multiple Prior Information

Huynh Van Luong, Nikos Deligiannis, Jürgen Seiler, Søren Forchhammer, André Kaup

    Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

    Abstract

    Online Robust Principle Component Analysis (RPCA) arises naturallyin time-varying signal decomposition problems such as videoforeground-background separation. We propose a compressive online RPCA algorithm that decomposes recursively a sequence of datavectors (e.g., frames) into sparse and low-rank components. Unlike conventional batch RPCA, which processes all the data directly, our method considers a small set of measurements taken per data vector (frame). Moreover, our method incorporates multiple prior information signals, namely previous reconstructed frames, to improve these paration and thereafter, update the prior information for the next frame. Using experiments on synthetic data, we evaluate the separation performance of the proposed algorithm. In addition, we apply the proposed algorithm to online video foreground and background separation from compressive measurements. The results show that the proposed method outperforms the existing methods.
    Original languageEnglish
    Title of host publicationProceedings of 2017 IEEE Global Conference on Signal and Information Processing (GlobalSIP)
    Number of pages5
    PublisherIEEE
    Publication date2017
    Pages1260-1264
    ISBN (Electronic)978-1-5090-5990-4
    DOIs
    Publication statusPublished - 2017
    Event5th IEEE Global Conference on Signal and Information Processing - Montreal, Canada
    Duration: 14 Nov 201716 Nov 2017
    Conference number: 5th

    Conference

    Conference5th IEEE Global Conference on Signal and Information Processing
    Number5th
    Country/TerritoryCanada
    CityMontreal
    Period14/11/201716/11/2017

    Keywords

    • Prior information
    • Robust PCA
    • n-1 minimization
    • Compressive measurements
    • Source separation

    Fingerprint

    Dive into the research topics of 'Compressive Online Robust Principal Component Analysis with Multiple Prior Information'. Together they form a unique fingerprint.

    Cite this