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

TL;DR

RAVQ-HoloNet introduces a rate-adaptive vector quantization framework for hologram compression, enabling high-quality reconstructions at various bit rates, significantly outperforming existing methods in rate-distortion performance.

Contribution

It presents the first rate-adaptive deep learning framework for hologram compression, allowing flexible bit rate control within a single network.

Findings

Outperforms state-of-the-art methods in BD-Rate by -33.91%.

Achieves BD-PSNR of 1.02 dB improvement at low bit rates.

Provides high-fidelity hologram reconstructions across different bit rates.

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.