Entropy-constrained tree-structured vector quantizer design

Kenneth Rose, David Jonathan Miller, Allen Gersho

    Research output: Contribution to journalArticle

    7 Citations (Scopus)

    Abstract

    Current methods for the design of pruned or unbalanced tree-structured vector quantizers such as the Generalized Breiman-Friedman-Olshen-Stone (GBFOS) algorithm are effective, but suffer from several shortcomings. We identify and clarify issues of suboptimality including greedy growing, the suboptimal encoding rule, and the need for time sharing between quantizers to achieve arbitrary rates. We then present the leaf-optimal tree design (LOTD) method which, with a modest increase in design complexity, alters and reoptimizes tree structures obtained from conventional procedures. There are two main advantages over existing methods. First, the optimal entropy-constrained nearest-neighbor rule is used for encoding at the leaves; second, explicit quantizer solutions are obtained at all rates without recourse to time sharing. We show that performance improvement is theoretically guaranteed. Simulation results for image coding demonstrate that close to 1 dB reduction of distortion for a given rate can be achieved by this technique relative to the GBFOS method.

    Original languageEnglish (US)
    Pages (from-to)393-398
    Number of pages6
    JournalIEEE Transactions on Image Processing
    Volume5
    Issue number2
    DOIs
    StatePublished - Dec 1 1996

    Fingerprint

    Entropy
    Image coding

    All Science Journal Classification (ASJC) codes

    • Software
    • Computer Graphics and Computer-Aided Design

    Cite this

    @article{ffe60553e68d4322abb94628270dc7f3,
    title = "Entropy-constrained tree-structured vector quantizer design",
    abstract = "Current methods for the design of pruned or unbalanced tree-structured vector quantizers such as the Generalized Breiman-Friedman-Olshen-Stone (GBFOS) algorithm are effective, but suffer from several shortcomings. We identify and clarify issues of suboptimality including greedy growing, the suboptimal encoding rule, and the need for time sharing between quantizers to achieve arbitrary rates. We then present the leaf-optimal tree design (LOTD) method which, with a modest increase in design complexity, alters and reoptimizes tree structures obtained from conventional procedures. There are two main advantages over existing methods. First, the optimal entropy-constrained nearest-neighbor rule is used for encoding at the leaves; second, explicit quantizer solutions are obtained at all rates without recourse to time sharing. We show that performance improvement is theoretically guaranteed. Simulation results for image coding demonstrate that close to 1 dB reduction of distortion for a given rate can be achieved by this technique relative to the GBFOS method.",
    author = "Kenneth Rose and Miller, {David Jonathan} and Allen Gersho",
    year = "1996",
    month = "12",
    day = "1",
    doi = "10.1109/83.480777",
    language = "English (US)",
    volume = "5",
    pages = "393--398",
    journal = "IEEE Transactions on Image Processing",
    issn = "1057-7149",
    publisher = "Institute of Electrical and Electronics Engineers Inc.",
    number = "2",

    }

    Entropy-constrained tree-structured vector quantizer design. / Rose, Kenneth; Miller, David Jonathan; Gersho, Allen.

    In: IEEE Transactions on Image Processing, Vol. 5, No. 2, 01.12.1996, p. 393-398.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - Entropy-constrained tree-structured vector quantizer design

    AU - Rose, Kenneth

    AU - Miller, David Jonathan

    AU - Gersho, Allen

    PY - 1996/12/1

    Y1 - 1996/12/1

    N2 - Current methods for the design of pruned or unbalanced tree-structured vector quantizers such as the Generalized Breiman-Friedman-Olshen-Stone (GBFOS) algorithm are effective, but suffer from several shortcomings. We identify and clarify issues of suboptimality including greedy growing, the suboptimal encoding rule, and the need for time sharing between quantizers to achieve arbitrary rates. We then present the leaf-optimal tree design (LOTD) method which, with a modest increase in design complexity, alters and reoptimizes tree structures obtained from conventional procedures. There are two main advantages over existing methods. First, the optimal entropy-constrained nearest-neighbor rule is used for encoding at the leaves; second, explicit quantizer solutions are obtained at all rates without recourse to time sharing. We show that performance improvement is theoretically guaranteed. Simulation results for image coding demonstrate that close to 1 dB reduction of distortion for a given rate can be achieved by this technique relative to the GBFOS method.

    AB - Current methods for the design of pruned or unbalanced tree-structured vector quantizers such as the Generalized Breiman-Friedman-Olshen-Stone (GBFOS) algorithm are effective, but suffer from several shortcomings. We identify and clarify issues of suboptimality including greedy growing, the suboptimal encoding rule, and the need for time sharing between quantizers to achieve arbitrary rates. We then present the leaf-optimal tree design (LOTD) method which, with a modest increase in design complexity, alters and reoptimizes tree structures obtained from conventional procedures. There are two main advantages over existing methods. First, the optimal entropy-constrained nearest-neighbor rule is used for encoding at the leaves; second, explicit quantizer solutions are obtained at all rates without recourse to time sharing. We show that performance improvement is theoretically guaranteed. Simulation results for image coding demonstrate that close to 1 dB reduction of distortion for a given rate can be achieved by this technique relative to the GBFOS method.

    UR - http://www.scopus.com/inward/record.url?scp=0030086415&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=0030086415&partnerID=8YFLogxK

    U2 - 10.1109/83.480777

    DO - 10.1109/83.480777

    M3 - Article

    VL - 5

    SP - 393

    EP - 398

    JO - IEEE Transactions on Image Processing

    JF - IEEE Transactions on Image Processing

    SN - 1057-7149

    IS - 2

    ER -