Three-dimensional image authentication with double random phase encryption in one capture

We propose a new, to the best of our knowledge, three-dimensional (3D) authentication method based on double random phase integral imaging, and only one image at one viewpoint is used during the authentication process. Two neural networks are applied to estimate depth information and the inpainted synthesized viewpoint image. The usage of deep learning and geometric refocusing techniques greatly simplifies the whole authentication process including capture, transmission, and storage. Experimental results verify the presented method and successfully prove the developed 3D authentication process using a nonlinear correlation method.