Abstract: The identification of image acquisition sources is an important problem in digital image forensics. This paper introduces a new feature-based method for digital camera identification. The method, which is based on an analysis of the imaging pipeline and digital camera processing operations, employs bi-coherence and wavelet coefficient features extracted from digital images. The sequential forward feature selection algorithm is used to select features, and a support vector machine is used as the classifier for source camera identification. Experiments indicate that the source camera identification method based on bi-coherence and wavelet coefficient features is both efficient and reliable.

@inproceedings{Meng:2008aa,
  publisher    = {Springer Boston},
  author       = {Fanjie Meng and Xiangwei Kong and Xingang You},
  url          = {http://www.springerlink.com/content/tm40245266808516/fulltext.pdf},
  series       = {IFIP International Federation for Information Processing},
  pages        = {207--218},
  volume       = {285},
  year         = {2008},
  editor       = {Indrajit Ray and Sujeet Shenoi},
  address      = {Kyoto, Japan},
  title        = {A new feature-based method for source camera identification},
  booktitle    = {Advances in Digital Forensics IV},
}