English

Unfolding Framework with Complex-Valued Deformable Attention for High-Quality Computer-Generated Hologram Generation

Computer Vision and Pattern Recognition 2025-09-01 v1

Abstract

Computer-generated holography (CGH) has gained wide attention with deep learning-based algorithms. However, due to its nonlinear and ill-posed nature, challenges remain in achieving accurate and stable reconstruction. Specifically, (ii) the widely used end-to-end networks treat the reconstruction model as a black box, ignoring underlying physical relationships, which reduces interpretability and flexibility. (iiii) CNN-based CGH algorithms have limited receptive fields, hindering their ability to capture long-range dependencies and global context. (iiiiii) Angular spectrum method (ASM)-based models are constrained to finite near-fields.In this paper, we propose a Deep Unfolding Network (DUN) that decomposes gradient descent into two modules: an adaptive bandwidth-preserving model (ABPM) and a phase-domain complex-valued denoiser (PCD), providing more flexibility. ABPM allows for wider working distances compared to ASM-based methods. At the same time, PCD leverages its complex-valued deformable self-attention module to capture global features and enhance performance, achieving a PSNR over 35 dB. Experiments on simulated and real data show state-of-the-art results.

Keywords

Cite

@article{arxiv.2508.21657,
  title  = {Unfolding Framework with Complex-Valued Deformable Attention for High-Quality Computer-Generated Hologram Generation},
  author = {Haomiao Zhang and Zhangyuan Li and Yanling Piao and Zhi Li and Xiaodong Wang and Miao Cao and Xiongfei Su and Qiang Song and Xin Yuan},
  journal= {arXiv preprint arXiv:2508.21657},
  year   = {2025}
}
R2 v1 2026-07-01T05:12:17.375Z