Improvement of image quality in wavefront-coding-reconstructed images under different apodization conditions using deep learning

Wavefront Coding is a powerful technology for extending depth of field, but it has the problem of generating artifacts such as ringing in reconstructed images. As countermeasures, the application of apodization and optimization using deep learning have been proposed. In this paper, we modified the apodization conditions of the intermediate images and reconstruction filters to generate the reconstructed images. We then reduced artifacts in the reconstructed images using deep learning with a high-resolution dataset, resulting in improved spatial frequency characteristics in 4K-equivalent high-resolution images.