Lee Siwoo, Nam Seung-Woo, Lee Juhyun, Jeong Yoonchan, Lee Byoungho
Opt Express. 2024 Mar 25;32(7):11107-11122. doi: 10.1364/OE.516564.
This study presents HoloSR, a novel deep learning-based super-resolution approach designed to produce high-resolution computer-generated holograms from low-resolution RGBD images, enabling the real-time production of realistic three-dimensional images. The HoloSR combines the enhanced deep super-resolution network with resize and convolution layers, facilitating the direct generation of high-resolution computer-generated holograms without requiring additional interpolation. Various upscaling scales, extending up to ×4, are evaluated to assess the performance of our method. Quantitative metrics such as structural similarity and peak signal-to-noise ratio are employed to measure the quality of the reconstructed images. Our simulation and experimental results demonstrate that HoloSR successfully achieves super-resolution by generating high-resolution holograms from low-resolution RGBD inputs with supervised and unsupervised learning.
本研究提出了HoloSR，这是一种基于深度学习的新型超分辨率方法，旨在从低分辨率RGBD图像生成高分辨率计算机生成全息图，从而实现逼真三维图像的实时生成。HoloSR将增强型深度超分辨率网络与调整大小和卷积层相结合，便于直接生成高分辨率计算机生成全息图，而无需额外的插值。评估了高达×4的各种放大比例，以评估我们方法的性能。采用结构相似性和峰值信噪比等定量指标来衡量重建图像的质量。我们的模拟和实验结果表明，HoloSR通过有监督和无监督学习从低分辨率RGBD输入生成高分辨率全息图，成功实现了超分辨率。
