High-quality computer-generated holography based on Vision Mamba

•The paper proposed a lightweight model, the CVMNet, based on the ViM for generating high-quality holograms while ensuring real-time performance. The CVMNet leverages the local feature extraction capabilities of convolutional layers and the long-range modeling abilities of state-space models (SSMs)...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Optics and lasers in engineering Jg. 184; S. 108704
Hauptverfasser: Yang, Lei, Xu, Shengyuan, Yang, Chunzheng, Chang, Chenliang, Hou, Qichao, Song, Qiang
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier Ltd 01.01.2025
Schlagworte:
Computer-generated holography
Deep learning
Long-range modeling
Parallel computation
Vision mamba
Deep learning
Vision mamba
Parallel computation
Long-range modeling
Computer-generated holography
ISSN:0143-8166
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:•The paper proposed a lightweight model, the CVMNet, based on the ViM for generating high-quality holograms while ensuring real-time performance. The CVMNet leverages the local feature extraction capabilities of convolutional layers and the long-range modeling abilities of state-space models (SSMs) to enhance computer-generated hologram (CGH) quality.•Compared to existing models, the CVMNet implements parallel computation for the ViM, effectively reducing the computational load and significantly decreasing the number of parameters.•Numerical reconstruction and optical experiments demonstrate that the CVMNet can generate 1080 P high-quality holograms in just 16 ms, boosting an average PSNR of over 30 dB and effectively suppressing speckle noise in reconstructed images. Generalization experiments demonstrate that the CVMNet has good generalization capabilities. Furthermore, both simulation and optical reconstruction of color holography using the proposed model are presented. The results show in our method, details are relatively crisp and colors are well-defined. Deep learning, especially through model-driven unsupervised networks, offers a novel approach for efficient computer-generated hologram (CGH) generation. However, current model-driven CGH generation models are primarily built on the convolutional neural networks (CNNs), which struggle to achieve high-quality hologram reconstruction due to limited receptive fields. Although Vision Transformers (ViTs) excel at processing more distant visual information, they are burdened with huge computational load. The recent emergence of Vision Mamba (ViM) presents a promising avenue to address these challenges. In this study, we introduce the CVMNet, a lightweight model that combines the precision of convolutional layers for local feature extraction and the long-range modeling abilities of state-space models (SSMs) to enhance the quality of CGHs. By employing parallel computation for the ViM to handle feature channels, the CVMNet effectively reduces the number of model parameters. Numerical reconstruction and optical experiments demonstrate that the CVMNet can generate 1080P high-quality holograms in just 16 ms, boosting an average PSNR of over 30 dB and effectively suppressing speckle noise in reconstructed images. Additionally, the CVMNet showcases robust generalization capabilities.
ISSN:0143-8166
DOI:10.1016/j.optlaseng.2024.108704