ACTIVE IMAGE AUTHENTICATION AND FORGERY LOCALIZATION FOR MOBILE DEVICES.

The paper presents a method for authenticating images that were shot using a mobile phone. Thanks to the newest advances in integrated cameras, smartphones represent one of the most popular ways to take a photograph. A picture could be critical evidence in a trial and, with the abundance of free editing software, a forgery is easy to achieve. A method to secure and test photographs for authenticity was developed, implemented as a mobile application, tested and detailed in this paper. The watermarking algorithm is based on an enhanced quantization index modulation (QIM) approach in the discrete cosine transform (DCT) domain and obtains very good perceptual quality of the watermarked images, with no visible artifacts to the naked eye. The watermark is also unaffected by recompression using quality factors greater than an adjustable preset value the algorithm is protected against. In order to further increase the security, the embedded watermark is encrypted using a private key. At the detection side, the mobile application allows the user not only to test the authenticity of an image, but also to precisely detect the forged image areas. Besides the watermark embedding and detection algorithms, the paper also includes test results, limits and optimizations for implementation on Android mobile devices. [ABSTRACT FROM AUTHOR]