In this work, we present the integration of the photon counting technique (PhCT) with an encryption system in the Gyrator domain (GD) for secure image authentication. The encryption system uses two random phase masks (RPMs), one RPM is defined at the spatial domain and the other RPM is defined at the GD, in order to encode the image to encrypt (original image) into random noise. The rotation angle of the Gyrator transform adds a new key that increases the security of the encryption system. The decryption system is an inverse system with respect to the encryption system. The PhCT limits the information content of an image in a nonlinear, random and controlled way; the photon-limited image only has a few pixels of information, this type of image is usually known as sparse image. We apply the PhCT over the encrypted image. The resulting image in the decryption system is not a copy of the original image, this decrypted image is a random code that should contain the sufficient information for the authentication of the original image using a nonlinear correlation technique. Finally, we evaluate the peak-to-correlation energy metric for different values of the parameters involved in the encryption and authentication systems, in order to test the verification capability of the authentication system.