Fig. 3: Simulation setup and results.

a Simulation setup. An image “Chelsea” from the scikit-image dataset²⁷ is convolved with a Prewitt operator (vertical edge detection). We explore the noise tolerance of both the analog and the hybrid optical computing systems by adding additive white Gaussian noise to the weights and examining the system’s performance by investigating the noise distribution of the outputs. b performance of the analog and hybrid computing schemes in terms of RMSE with different SNRs. The following results are obtained at an SNR of 25 dB. c, f Processed and reconstructed images by the analog and hybrid computing systems, respectively. d, g Distribution of expected pixel values against the processed pixel values (both normalized), for the analog and hybrid computing systems, respectively. Insets show the corresponding processed images. Noisy pixels can be clearly observed in the image processed using analog computing. e, h Noise distribution of the analog and hybrid computing systems, respectively. Analog computing reveals a Gaussian noise distribution with a standard deviation of 0.027, corresponding to a numerical precision of 3.6 bits. The HOP shows a greatly improved noise distribution thanks to the introduction of logic levels and decisions based on thresholding.