Double random phase encoding with a Poisson-multinomial distribution for efficient colorful image authentication

Abstract

In this paper, a new integrated approach is proposed for authenticating digital color image. Here, a new technique, Poisson-Multinomial Distribution (PMD) is introduced for the first time in image processing. It is used for Photon Counting Imaging (PCI) which is integrated with Double Random Phase Encoding (DRPE) scheme in this system. The main goal of this proposal is to establish an image authentication architecture which will only work for 3D digital color images. It is a simplified way for applying PCI scheme on three channels of a 3D image simultaneously. The system will generate a stationary white noise as a final output, which will be difficult to decode for a third party attacker. This proposed scheme works directly with the original digital RGB image. At first, the system encrypts those three channels of the image individually with DRPE method without separating them and the amplitude part of the encrypted image is photon counted using PMD. Finally, to obtain an optimal result, a probability density function is used. On the authentication part, the reference digital image is encrypted by the same keys as the original image and then both of the encrypted images are compared with a statistical nonlinear correlation method. The numerical experiments say that, this proposed PMD based method is proven to be a good and simplified one that can be used to encrypt digital color images. In addition, even if the number of photons is really low, this new system can successfully differentiate between true class and false class images. To prove its efficiency we have also shown some experimental results under different situations.