Journal of Applied Sciences ›› 2014, Vol. 32 ›› Issue (6): 596-604.doi: 10.3969/j.issn.0255-8297.2014.06.008

• Signal and Information Processing • Previous Articles     Next Articles

Detecting Compression Traces in Multiple JPEG-Compressed Image

Han Hong-li1, LI Ye-zhou1, NIU Shao-zhang2, SUN Xiao-ting1   

  1. 1. School of Science, Beijing University of Posts and Telecommunications, Beijing 100876, China
    2. Beijing Key Lab of Intelligent Telecommunication Software and Multimedia, Beijing University of Posts and Telecommunications, Beijing 100876, China
  • Received:2014-03-23 Revised:2014-08-07 Online:2014-11-28 Published:2014-08-07

Abstract: Although some tamper detection methods based on compression characteristics work well for
tampered images that experienced double JPEG compression, they often fail for multiple JPEG compression.
That is because multiple JPEG compression may significantly change the DCT coefficient statistics of the
tampered images. Aimed at low quality network pictures that may be tampered, it is shown that approximate
recovery of the image may be done with a compression-removing method. This is feasible if one can explore
the compression trace of the image that has been compressed with various quality factors. This paper presents
a method to detect compression traces of images by calculating the quantization error of DC coefficient in the
DCT domain. A compression-removing method is then used to approximately recover the image. The same
method is used to detect compression traces in the recovered image afterwards. All compression traces can be
found by repeating the process. If there are different compression traces in different parts of the image, tamper
detection is achieved.

Key words: multiple JPEG compression, compression trace, quantization error

CLC Number: