Denoising of Fourier microscopic images based on attention mechanism and parallel residuals

To solve the problem that the quality of the reconstructed image is degraded due to the influence of noise in the captured image during Fourier ptychographic microscopy, an attention mechanisms and parallel residual denoising network (AMCPN) of Fourier ptychographic microscopy images based on attention mechanism and parallel residual residuals is proposed.The network serially processes the channel and spatial attention mechanism, integrates the feature information of different channels and spaces, reduces the influence of noise and improves the robustness of the denoising network.A parallel residual block structure is designed, and the input features are extracted through two parallel branches, which fully excavates useful features, increases the number of channels in the last layer of each branch, and improves the network's ability to retain image detail information. Experimental results show that the peak signal-to-noise ratio (PSNR) and structure similarity (SSIM) values of the reconstructed image quality are increased by more than 2.87 dB and 0.088 6, respectively, under different intensity noises, which can effectively remove the image noise, improve the quality of the reconstructed image, and ensure the high spatial resolution and clarity of the reconstructed image.