Speckle noise reduction for digital holographic images using Swin Transformer

We introduce an innovative approach for reducing speckle noise in holographic reconstruction images utilizing the Transformer architecture. This approach not only effectively captures speckle noise from digital holographic images but also better preserves details in images, owing to the characteristics of the Swin Transformer in globally and locally capturing relationships between image features. The network is trained using a large dataset with a distribution similar to real speckle noise. Experimental results demonstrate outstanding denoising performance of the proposed method and effectively preserving the details.