On the Security of the Visual Hash Function

Regunathan Radhakrishnan, Ziyou Xiong, Nasir Memon

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    Robust hash functions are central to the security of multimedia content authentication systems. Such functions are sensitive to a key but robust to many allowed signal processing operations on the underlying content. Robustness of the hash function to changes in the original content implies the existence of a cluster in the feature space around the original contents feature vector, any point within which getting hashed to the same output. The shape and size of the cluster determines the trade-off between the robustness offered and the security of the authentication system based on the robust hash function. The clustering itself is based on a secret key and hence unknown to the attacker. However, we show in this paper that the specific clustering arrived at by a robust hash function may be possible to learn. Specifically, we look at a well known robust hash function for image data called the Visual Hash Function (VHF). Given just an input and its hash value, we show how to construct a statistical model of the hash function, without any knowledge of the secret key used to compute the hash. We also show how to use this model to engineer arbitrary and malicious collisions. Finally, we propose one possible modification to VHF so that constructing a model that mimics its behavior becomes difficult.

    Original languageEnglish (US)
    Title of host publicationProceedings of SPIE - The International Society for Optical Engineering
    EditorsE.J. Delp III, P. Wah Wong
    Pages644-652
    Number of pages9
    Volume5020
    DOIs
    StatePublished - 2003
    EventSecurity and Watermarking of Multimedia Contents V - Santa Clara, CA, United States
    Duration: Jan 21 2003Jan 24 2003

    Other

    OtherSecurity and Watermarking of Multimedia Contents V
    CountryUnited States
    CitySanta Clara, CA
    Period1/21/031/24/03

    Fingerprint

    Hash functions
    Authentication
    multimedia
    Signal processing
    engineers
    signal processing
    Engineers
    collisions
    output

    Keywords

    • Boosting
    • Hash function
    • Multimedia Content Authentication
    • Visual Hash
    • Weak Collision Property

    ASJC Scopus subject areas

    • Electrical and Electronic Engineering
    • Condensed Matter Physics

    Cite this

    Radhakrishnan, R., Xiong, Z., & Memon, N. (2003). On the Security of the Visual Hash Function. In E. J. Delp III, & P. Wah Wong (Eds.), Proceedings of SPIE - The International Society for Optical Engineering (Vol. 5020, pp. 644-652) https://doi.org/10.1117/12.477340

    On the Security of the Visual Hash Function. / Radhakrishnan, Regunathan; Xiong, Ziyou; Memon, Nasir.

    Proceedings of SPIE - The International Society for Optical Engineering. ed. / E.J. Delp III; P. Wah Wong. Vol. 5020 2003. p. 644-652.

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Radhakrishnan, R, Xiong, Z & Memon, N 2003, On the Security of the Visual Hash Function. in EJ Delp III & P Wah Wong (eds), Proceedings of SPIE - The International Society for Optical Engineering. vol. 5020, pp. 644-652, Security and Watermarking of Multimedia Contents V, Santa Clara, CA, United States, 1/21/03. https://doi.org/10.1117/12.477340
    Radhakrishnan R, Xiong Z, Memon N. On the Security of the Visual Hash Function. In Delp III EJ, Wah Wong P, editors, Proceedings of SPIE - The International Society for Optical Engineering. Vol. 5020. 2003. p. 644-652 https://doi.org/10.1117/12.477340
    Radhakrishnan, Regunathan ; Xiong, Ziyou ; Memon, Nasir. / On the Security of the Visual Hash Function. Proceedings of SPIE - The International Society for Optical Engineering. editor / E.J. Delp III ; P. Wah Wong. Vol. 5020 2003. pp. 644-652
    @inproceedings{8e714f9e049246ada1bda9be1504723f,
    title = "On the Security of the Visual Hash Function",
    abstract = "Robust hash functions are central to the security of multimedia content authentication systems. Such functions are sensitive to a key but robust to many allowed signal processing operations on the underlying content. Robustness of the hash function to changes in the original content implies the existence of a cluster in the feature space around the original contents feature vector, any point within which getting hashed to the same output. The shape and size of the cluster determines the trade-off between the robustness offered and the security of the authentication system based on the robust hash function. The clustering itself is based on a secret key and hence unknown to the attacker. However, we show in this paper that the specific clustering arrived at by a robust hash function may be possible to learn. Specifically, we look at a well known robust hash function for image data called the Visual Hash Function (VHF). Given just an input and its hash value, we show how to construct a statistical model of the hash function, without any knowledge of the secret key used to compute the hash. We also show how to use this model to engineer arbitrary and malicious collisions. Finally, we propose one possible modification to VHF so that constructing a model that mimics its behavior becomes difficult.",
    keywords = "Boosting, Hash function, Multimedia Content Authentication, Visual Hash, Weak Collision Property",
    author = "Regunathan Radhakrishnan and Ziyou Xiong and Nasir Memon",
    year = "2003",
    doi = "10.1117/12.477340",
    language = "English (US)",
    volume = "5020",
    pages = "644--652",
    editor = "{Delp III}, E.J. and {Wah Wong}, P.",
    booktitle = "Proceedings of SPIE - The International Society for Optical Engineering",

    }

    TY - GEN

    T1 - On the Security of the Visual Hash Function

    AU - Radhakrishnan, Regunathan

    AU - Xiong, Ziyou

    AU - Memon, Nasir

    PY - 2003

    Y1 - 2003

    N2 - Robust hash functions are central to the security of multimedia content authentication systems. Such functions are sensitive to a key but robust to many allowed signal processing operations on the underlying content. Robustness of the hash function to changes in the original content implies the existence of a cluster in the feature space around the original contents feature vector, any point within which getting hashed to the same output. The shape and size of the cluster determines the trade-off between the robustness offered and the security of the authentication system based on the robust hash function. The clustering itself is based on a secret key and hence unknown to the attacker. However, we show in this paper that the specific clustering arrived at by a robust hash function may be possible to learn. Specifically, we look at a well known robust hash function for image data called the Visual Hash Function (VHF). Given just an input and its hash value, we show how to construct a statistical model of the hash function, without any knowledge of the secret key used to compute the hash. We also show how to use this model to engineer arbitrary and malicious collisions. Finally, we propose one possible modification to VHF so that constructing a model that mimics its behavior becomes difficult.

    AB - Robust hash functions are central to the security of multimedia content authentication systems. Such functions are sensitive to a key but robust to many allowed signal processing operations on the underlying content. Robustness of the hash function to changes in the original content implies the existence of a cluster in the feature space around the original contents feature vector, any point within which getting hashed to the same output. The shape and size of the cluster determines the trade-off between the robustness offered and the security of the authentication system based on the robust hash function. The clustering itself is based on a secret key and hence unknown to the attacker. However, we show in this paper that the specific clustering arrived at by a robust hash function may be possible to learn. Specifically, we look at a well known robust hash function for image data called the Visual Hash Function (VHF). Given just an input and its hash value, we show how to construct a statistical model of the hash function, without any knowledge of the secret key used to compute the hash. We also show how to use this model to engineer arbitrary and malicious collisions. Finally, we propose one possible modification to VHF so that constructing a model that mimics its behavior becomes difficult.

    KW - Boosting

    KW - Hash function

    KW - Multimedia Content Authentication

    KW - Visual Hash

    KW - Weak Collision Property

    UR - http://www.scopus.com/inward/record.url?scp=0242576290&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=0242576290&partnerID=8YFLogxK

    U2 - 10.1117/12.477340

    DO - 10.1117/12.477340

    M3 - Conference contribution

    VL - 5020

    SP - 644

    EP - 652

    BT - Proceedings of SPIE - The International Society for Optical Engineering

    A2 - Delp III, E.J.

    A2 - Wah Wong, P.

    ER -