Image Tamper Detection using Location Decision Embedding Technique

Abstract

The fast and dynamic growth in technology erodes the trust in the integrity of imagery. With the wide availability of digital technology and tools a few affected areas with respect to cyber world are sophisticated photo hoaxes in media, political campaigns, armed forces, fashion and entertainment industry. The rapid rate at which the digital crimes are increasing has become a major concern, especially in the court of law. To cater these needs digital signatures were used, but they are no longer considered as a proof of authenticity or integrity, as there are tools which generate a valid digital signature of tampered images that pass through the validation of image authentication software of reputed digital cameras. To verify the integrity of an image, an authentication mechanism is introduced in this paper by improvising the image acquisition model as a possible solution to cater the needs of legal vindications. In this approach a Verification code of the image is generated using Gödelization technique and embedded in the image using LDET which acts as in-camera finger prints (watermark) of the image to detect tampering if any. The tampered area is localised by detecting and extracting invariant features of the image using SURF, a key point descriptor and matching these features with Euclidean distance. The results prove that the Improvised Image Acquisition Model detects Image tampering and Tamper Localization method locates copy move tampered regions which is robust to scaling and rotation transformations.