Systematic Degradation Analysis in Phase-Only Computational Holography: A Simulation Framework

Résumé fourni par la source

Phase-only spatial light modulators (SLMs) are the dominant hardware substrate for real-time holographic display, but the practical trade-offs between SLM hardware constraints and reconstruction quality remain incompletely characterised in a systematic, reproducible way. We present HOLOFORGE, an open-source Python simulation framework that isolates four independently controllable degradation axes—spatial resolution, phase quantisation bit-depth, viewing-angle bandwidth, and coherent speckle noise—and measures their impact on holographic reconstruction quality using three complementary metrics: PSNR, SSIM, and LPIPS. All degradations are applied in the physically correct domain: resolution and bandwidth limits act on the complex SLM field, and speckle is modelled as random aperture-plane phase noise prior to coherent propagation rather than as additive intensity noise. Across parameter sweeps on four synthetic test scenes and five random phase-retrieval seeds, we identify a sharp, scene- and seed-robust quality cliff at 1-bit phase quantisation (SSIM = 0.92 at 2 bits vs. SSIM = 0.003 at 1 bit on the reference scene; 0.59 → 0.01 averaged across scenes), nearlossless reconstruction of low-spatial-frequency content down to 25% viewing-angle bandwidth (SSIM > 0.97), and a systematic divergence between PSNR and SSIM as degradation increases. The framework, experiment scripts, and raw result data are released at https://github.com/ultramagnus23/HoloForge as a reproducible baseline for Part 2, which will establish perceptual detection thresholds via a human-observer study

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.