Abstract: JPEG is probably the most widely used image compression standard in taking digital pictures, e.g., in most digital cameras. As a result, synthetic images by the trick operation of copy-paste are usually from and to JPEG images. Realizing that it might be impossible to find a method that is universal for all kinds of forgeries, we proposed a novel blind approach to detect copy-paste trail in doctored JPEG images and meanwhile locate the doctored area. The approach works well even when a JPEG image is truncated or multi-compressed, by extract the DCT block artifact grid and detect mismatch of the grid. Experiments well demonstrate the effectiveness of the proposed approach. However, most authentication algorithms require pending image to be uncompressed and high quality. As for the widely used JPEG format, only two kinds of algorithms were proposed. One is used to detect double JPEG compression. It is already noticed [2] that periodic property caused by double quantization appears in the histogram of DCT coefficient. An algorithm [5] was proposed to detect doctored JPEG images and further locate the doctored parts by examine the double quantization effect hidden among the DCT coefficients. A serious matter is that the detection will fail if the image is thrice compressed.

  author       = {Weihai Li and Yuan Yuan and Nenghai Yu},
  url          = {http://www.eurasip.org/Proceedings/Ext/LNLA2008/papers/cr1006.pdf},
  title        = {Detecting copy-paste forgery of {JPEG} image via block artifact grid extraction},
  year         = {2008},
  keywords     = {Image Forensics; Tamper Detection; JPEG},
  booktitle    = {Proceedings of the 2008 International Workshop on Local and Non-Local Approximation in Image Processing},