Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
456280 | Digital Investigation | 2012 | 11 Pages |
In this modern age in which we are living, digital images play a vital role in many application areas. But at the same time the image retouching techniques has also increased which forms a serious threat to the security of digital images. To cope with this problem, the field of digital forensics and investigation has emerged and provided some trust in digital images. In this paper we proposed a technique for image authentication that detects the manipulations that are done in the digital images. In most of the image forgeries such as copy-and-paste forgery, region duplication forgery, image splicing forgery etc basic image operations or manipulations are often involved. Thus if there exists the evidence for basic image alterations in digital images we can say that the image has been altered. This paper aims at detecting the basic image operations such as re-sampling (rotation, rescaling), contrast enhancement and histogram equalization which are often done in forged images. The available interpolation related spectral signature method is used for detecting rotation and rescaling and for estimating parameters such as rotation angle and rescale factors. This rotation/rescaling detection method detects some unaltered images as altered one when the images are JPEG compressed. We have overcome that problem by adding noise in the input images. We have also used the existing fingerprint detection technique for detecting contrast enhancement and histogram equalization. Besides the techniques discussed in the existing method, we identified a unique property for histogram equalization which can help us to differentiate contrast enhancement from histogram equalization. This work is tested in USC-SIPI database which consists of general unaltered images and achieved results with satisfactory accuracy.
► Detecting the re-sampling (rotation/rescaling) operation done in digital images. ► Rotation angle/Rescale factor is estimated in the detected forged image. ► Forgeries such as contrast enhancement and histogram equalization are detected. ► Add Gaussian noise to improve the re-sampling detection in JPEG compressed image.