A New Algorithm for High Dynamic Range Image Denoising based on Generative Adversarial Networks

Abstract

In this study, a new algorithm for high dynamic range image denoising based on generative adversarial network is successfully designed and implemented, and a series of innovative and practical research results are achieved in the field of image denoising. In terms of algorithm design, the structure of the generative adversarial network is innovatively optimised, which enhances the ability of retaining image details, judges the authenticity of the generated image more effectively, and improves the adversarial effect of the generative adversarial network. In terms of loss function design, a multi-loss function optimisation strategy that integrates the consideration of adversarial loss, content loss and structural loss is proposed to make the generated denoised image visually similar to the real noise-free image through the adversarial game between the generator and the discriminator, reasonably adjusting the weight of each loss term to achieve effective constraints on the generator and improve the quality of the denoised image. Compared with traditional denoising algorithms such as mean filtering, median filtering and Gaussian filtering, this algorithm has significant advantages in objective evaluation indexes such as peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM), which can remove noise more effectively while retaining the details and structural information of the image in a better way, and the visual effect of the denoised image is significantly improved. The denoising time of a single image in the testing stage is short, and the memory consumption is within the acceptable range, which can meet the needs of practical applications.