Tesi di LAUREA SPECIALISTICA Titolo Image tampering detection and localization Data 2014-07-21 Autore/i Gaborini, Lorenzo Relatore Tubaro, S. Full text non disponibile Abstract Image tampering is nowadays at everyone's reach: this has determined the urgent necessity of forensic tools capable of blindly distinguishing whether and where an image has been altered. To do so without having any prior information on the examined images reveals to be a challenging task. The goal of this work is to propose two algorithms respectively aimed at classifying whether any given image is fake or not, and locating such tampering. These algorithms are inspired by those who won the IEEE First Image Forensic Challenge, the first competition organized specifically to compare state-of-the-art methods on blind image forensics. In particular, the first algorithm is strongly based on a machine learning approach, while the second one is a fusion between two entirely novel techniques with another well-known method. Results, validated against the same dataset used in the Challenge, show that in both cases our proposed algorithms reach or beat the state-of-the-art performance on the specific dataset, yet leaving many questions open to discussion and further improvement. As a side effect of a deep joint analysis of the Challenge dataset along with the first proposed algorithm, we also uncovered a new forensic fingerprint which could be used for other attacks, proving also that it has been unknowingly exploited in the winning submission.