RAVQ-HoloNet: Rate-Adaptive Vector-Quantized Hologram Compression

2025-11-25

AI Summary

To address the critical bottleneck of high compression demand for holographic images in AR/VR applications and the lack of bit-rate adaptability in existing deep learning-based compression methods, this paper proposes an end-to-end rate-distortion-optimized compression framework. The core innovation is a rate-adaptive vector quantization (RA-VQ) mechanism, enabling flexible multi-bit-rate switching using a single neural network—thereby significantly improving reconstruction fidelity, especially at low and ultra-low bit rates. The framework integrates deep neural networks with differentiable vector quantization to support dynamic bit-rate adjustment without retraining. Experimental results on standard benchmark datasets demonstrate that our method achieves a 33.91% BD-Rate reduction and a 1.02 dB BD-PSNR gain over state-of-the-art approaches, establishing new benchmarks in both reconstruction quality and bit-rate flexibility.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Machine Learning: Learning on the Edge & Model Compression
• Search and Optimization: Non-convex Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Holography offers significant potential for AR/VR applications, yet its adoption is limited by the high demands of data compression. Existing deep learning approaches generally lack rate adaptivity within a single network. We present RAVQ-HoloNet, a rate-adaptive vector quantization framework that achieves high-fidelity reconstructions at low and ultra-low bit rates, outperforming current state-of-the-art methods. In low bit, our method exceeds by -33.91% in BD-Rate and achieves a BD-PSNR of 1.02 dB from the best existing method demonstrated by the rate-distortion curve.

Problem

• Compressing holographic data for AR/VR applications
• Achieving rate adaptivity within a single network
• Improving reconstruction fidelity at low bit rates

Innovation

• Rate-adaptive vector quantization framework for holograms
• Achieves high-fidelity reconstructions at low bit rates
• Outperforms state-of-the-art compression methods significantly