Fig. 1. Schematic illustration of the neuromorphic encryption and decryption processes.

Fig. 2. Neuromorphic encryption system. (a) The optical configuration. (b) An encrypted letter “N” is made with a diffuse metal surface, captured with frame-mode and a focusing lens. (c) An event camera is employed as the neuromorphic sensor to collect the encrypted data.

Fig. 3. Analysis of encryption-related data. (a) The accumulated speckle and corresponding retrieved autocorrelation and the final decrypted results in different scattering conditions are presented. (b) The normalized intensity of the dashed line in the accumulated speckle. (c) The normalized intensity of the dashed line in the retrieved autocorrelation.

Fig. 4. Simulation results. The virtual cyphertext of speckle patterns from intermediate processes is used for event-stream data generation. The accumulated speckles with encrypted events and decrypted images with the learning PR algorithm.

Fig. 5. Robustness test against cropping. Ciphertext with different cropping ratios are 1/16, 1/8, 1/4, and 1/2, respectively.

Fig. 6. Robustness test against noise. Ciphertext added zero-mean Gaussian white noise with variances of 0.01 and 0.02 and added salt and pepper noise with 0.01 and 0.02 noise densities.

Fig. 7. Experimental results. (a) Letter target “N.” (b) Letter target “L.” (c) Quick draw target “Tshirt.” (d) Quick draw target “Apple.” For each subfigure: the first column from the left is the ground truth plaintext; the first row is the collected encrypted events that have the same time period and different time periods; the second and third rows are the accumulated speckle and corresponding autocorrelation; the last row is the decrypted result via the learning PR algorithm.

Fig. 8. Comparison results of the traditional HIO PR algorithm with our learning method for event data decryption.

Fig. 9. Cracking resistance test with wrong time duration selection (in the red dash box). The correct process with a proper time duration selection is also presented in the green dash box.