• DocumentCode
    2827671
  • Title

    Secure JPEG steganography by LSB+ matching and multi-band embedding

  • Author

    Wu, Hao-Tian ; Huang, Jiwu

  • Author_Institution
    Sch. of Inf. Sci. & Technol., Sun Yat-Sen Univ., Guangzhou, China
  • fYear
    2011
  • fDate
    11-14 Sept. 2011
  • Firstpage
    2737
  • Lastpage
    2740
  • Abstract
    A new steganographic algorithm is proposed for JPEG images by modifying the block DCT coefficients. Firstly, an embedding algorithm called LSB+ matching is generated to approximately preserve the marginal distribution of DCT coefficients. We further divide the DCT coefficients into four frequency bands, including the direct current (DC), low-frequency, middle-frequency, and high-frequency. Via matrix encoding, low data hiding rate and high embedding efficiency are achieved in high-frequency band, while the hiding rate is increased in the middle-frequency and DC bands, and highest in the low-frequency band. In addition, a coefficient selection strategy is employed to make the hidden message less detectable. The proposed algorithm is implemented on a set of 10000 images and tested with four steganalytic algorithms. The experimental results show that it outperforms the F5, nsF5 and LSB+ algorithms in terms of detection accuracy for the JPEG images at various levels of quality.
  • Keywords
    discrete cosine transforms; image coding; image matching; steganography; JPEG images; LSB matching; block DCT coefficients; data hiding rate; discrete cosine transform; hidden message; marginal distribution; matrix encoding; multiband embedding; secure JPEG steganography; steganalytic algorithms; Accuracy; Approximation algorithms; Conferences; Discrete cosine transforms; Histograms; Image coding; Transform coding; JPEG stegnography; LSB+ matching; matrix encoding; multi-band embedding; steganalysis;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Image Processing (ICIP), 2011 18th IEEE International Conference on
  • Conference_Location
    Brussels
  • ISSN
    1522-4880
  • Print_ISBN
    978-1-4577-1304-0
  • Electronic_ISBN
    1522-4880
  • Type

    conf

  • DOI
    10.1109/ICIP.2011.6116235
  • Filename
    6116235