Passive And Blind Forensics Technology For Digital Image Tampering

Malicious tampering of digital images has a negative impact to people's social life. In theworst cases, it may damage the personal honor and the safety of people's lives and property, againstthe normal judicial process, or even threaten national security. Passive and Blind forensicstechnology is an effective method to fight against digital image tampering. It is an importantbranch of information security. Compared to the initiative forensics technology for digital imagetampering, the passive and blind forensics technology does not need to embed a watermark in theoriginal image. By analyzing the properties of the image acquisition device or the statisticalcharacteristics of the image directly, It can detect that whether the image is tampered or not.In this thesis, the types of digital image tampering and a variety of passive forensicstechnology are analyzed. According to the new digital image compression standard of JPEG2000,JPEG2000 codec, JPEG codec and the comparative between them are analyzed. Through thefrequency statistics characteristics of the discrete wavelet transform (DWT) coefficients inJPG2000 compression between single compression and double compression, and analyzing thecharacteristics of the JPEG2000 double compression, we achieved the following researches andinnovations.At the beginning, according to the perspective of those who tampered images, a newclassification is generalized aiming at the popular tampering means. The entire framework ofdigital image forensics technology are systematically summarized and generalized. The advantagesand disadvantages of various forensics technologies are detailed analyzed. At the request of thepractical application, the difficulty and the key of the forensics system are explained. Theguidelines of the standards samples databank of digital image testing for forensics are discussedand summarized.Secondly, JPEG2000 codec which is the new digital image compression standard is deeplyanalyzed. Aiming at the deficiencies and shortcomings of JPEG standard, the improvement andenhancement of JPEG2000 standard are summarized. The similarities and differences betweenJPEG and JPEG2000 standard are comparatively analyzed, and from this comparative analysis, the image statistical characteristics which can be used for image forensics in the new JPEG2000 imagecompression standard are outlined.Thirdly, the discrete wavelet transform (DWT) coefficients in JPEG2000 standard areresearched and analyzed. The frequency value about the sub-band coefficients of the discretewavelet transform is systematically counted and compared between the single compression and thedouble compression. We find that the statistical characteristics histogram about the discrete cosinetransform (DCT) frequency of the JPEG image compression standard are different from theJPEG2000 image compression standard between the single compression and the doublecompression. Through analyzing the "filling" feature of the JPEG2000 standards in the case ofdouble compression quantitatively, a method of "Compare of Field-Maximum-Value" is proposedwhich ban be used to detect double compression tampering of JPEG2000 format image.Finally, through analyzing the different compression characteristics between the tamperedblocks and non-tampered blocks in different images, we proposed a blind and passive forensicstechnology which uses the "filling" feature of the JPEG2000 standard in the case of doublecompression.