Use of sensor noise for source identification

A type of sensor noise known as photo response nonuniformity noise (PRNU) has been used in digital forensic applications. It can be considered as a fingerprint of the sensor which can uniquely identify each individual device. The reliability of the PRNU in device identification, however, depends on how accurate the noise is extracted from the digital data. If the noise is contaminated by the content of the digital data, the identification accuracy will drop. In this paper, we propose a novel wavelet-based denoising approach to reliably extract the PRNU. The denoising operation is applied adaptively so as to remove the scene content that is found in the noise. 10 cameras and 1133 samples are used to examine the effectiveness of the proposed method. Experimental results showed that the proposed method outperforms existing method in terms of identification accuracy and false acceptance rate.