PHOTO RESPONSE NON-UNIFORMITY BASED CAMERA IDENTIFICATION

Abstract

With the rapid development in the digital camera technology, a large amount of information carrying data in form of images are distributed through the internet. Digital camera identification which is defined as, the identification of source camera which captures an image, gaining its importance day by day. As camera identification is used in presenting the evidences in the courtroom and helping the legal investigation. In latest years, sensor pattern noise is becoming an attractive measure for the identification of the camera. Photo response non-uniformity (PRNU), caused by different sensitivity of the pixels towards the light, and is used for the identification of camera. However, the obtained PRNU is the mixture of various noises such as random noise, fixed pattern noise etc , JPEG compression and scene contents, which affects the stable identification. Moreover many previous methods fails in identification, when the query image is geometrically transformed. In this dissertation, we propose a novel digital camera identification method, using positive clustered sensor noise, which are robust to noise contamination. By experimental results, we show that the proposed method can identify the source camera from the query image (whether crop or uncropped image) with high accuracy.