Fig. 5
From: Image processing with Optical matrix vector multipliers implemented for encoding and decoding tasks
Application of the optical autoencoder as a DAE. a Schematic of the DAE training process, involving pairs of clean and noisy images to train the model. b Loss reduction over training epochs, demonstrating the effectiveness of the DAE in denoising input images. c Examples of noisy input and denoised output images, showing the improvement in image quality after denoising
