Zero-Order Diffraction Suppression in Full Field-of-View Computer Generated Holography: A Camera In the Loop Interferometric Approach

Technical Summary: Zero-Order Diffraction Suppression in Full Field-of-View Computer Generated Holography

Problem Statement
Phase-only computer-generated holography (CGH) is a critical technology for applications ranging from beam shaping to near-eye displays (NEDs) for augmented reality (AR). However, a persistent limitation in holographic displays is the presence of zero-order diffraction (ZOD). Due to the pixelated nature of spatial light modulators (SLMs) and imperfections such as unmodulated inter-pixel regions and imperfect anti-reflection coatings, a portion of incident light is reflected specularly. This creates an unwanted, bright central spot in the reconstructed virtual image.

Existing mitigation strategies have significant drawbacks:

• Spatial Filtering: Placing an obstruction in the Fourier plane (4f system) blocks the ZOD but requires bulky optics incompatible with miniaturized AR devices and creates a dark spot in the image center.
• Displacement: Shifting the ZOD laterally or axially reduces the effective field of view (FOV) or introduces aberrations and focus issues.
• Phase Compression: Reshaping phase profiles to minimize residual light has achieved only modest suppression (approx. 67%) with degraded diffraction efficiency.
• Interferometric Correction: Previous linear interferometric schemes have struggled with phase and amplitude mismatches, yielding low experimental suppression (approx. 32%) despite high theoretical potential.

Methodology
The authors propose a novel Camera-in-the-Loop (CITL) interferometric approach that suppresses ZOD via destructive interference directly within the phase encoding process, eliminating the need for external spatial filtering. The method relies on engineering the SLM's phase pattern to generate a "corrective beam" that matches the amplitude and has the exact opposite phase of the native ZOD.

The process involves four key stages:

1. Amplitude Estimation: An initial hologram is displayed that contributes zero intensity at the ZOD location. By measuring the total intensity and the residual ZOD intensity, the system calculates the ratio between the modulated image light and the native ZOD amplitude. This determines the necessary amplitude weighting for the corrective beam.
2. Uniform Phase Probing ("Piston Sweep"): A series of uniform phase maps are applied to the corrective beam component. A camera (Cam1), optically conjugated to the SLM plane through a 4f telescope with a pinhole (acting as a low-pass filter), records the interference intensity between the corrective beam and the native ZOD.
3. Pixel-Resolved Phase Retrieval: By analyzing the intensity response of each camera pixel across the different piston phases, the system identifies the specific phase value ($\phi_{i,min}$) that minimizes the ZOD intensity for that pixel. This generates a high-resolution, non-uniform phase map ($\Phi_{cam}$) that accounts for spatial variations in the native ZOD.
4. Affine Transformation and Iteration: The camera-derived phase map is mapped to the SLM's physical grid using an affine transformation to correct for geometric disparities (scaling, rotation, shear). To handle the periodic nature of phase (0 to $2\pi$), nearest-neighbor interpolation is used to avoid artifacts. This retrieved map serves as a baseline for further iterative "piston" sweeps, progressively reducing the discrepancy between the target and actual wavefront until convergence.

Once calibrated, this specific phase map can be applied to any subsequent hologram without recomputation, provided the amplitude weights are adjusted for the specific image content.

Key Contributions

• Full Field-of-View Operation: Unlike displacement methods, this technique suppresses the ZOD at the center of the image without sacrificing FOV or introducing comatic aberrations.
• Preservation of Modulation Depth: The method does not require phase compression, thereby preserving the full modulation depth and diffraction efficiency of the SLM.
• Robustness and Reusability: The corrective phase map is system-specific but hologram-agnostic. After a single calibration, the correction applies to arbitrary holograms, enabling real-time operation.
• Compact Implementation: The method removes the need for bulky 4f spatial filtering systems, making it suitable for compact, miniaturized AR/NED architectures.

Experimental Results
The authors validated the method using a 532 nm laser and a reflective LCoS SLM (Holoeye PLUTO-2.1).

• Suppression Efficiency: The technique achieved up to 99% suppression of the ZOD intensity (a two-order-of-magnitude reduction) in point-cloud and 2D holograms.
• Image Quality: The correction preserved image integrity. For point-cloud holograms, the peak-to-valley uniformity decreased only marginally (from 0.85 to 0.83). For complex tree holograms, image quality metrics (PSNR, MSE, SSIM) remained nearly identical between corrected and uncorrected states.
• 3D Performance: In multi-plane AR demonstrations, the method successfully eliminated the central haze and low-contrast artifacts associated with ZOD across different focal planes, enhancing the fidelity of 3D reconstructions.
• Computational Speed: Once the phase map is retrieved, applying the correction to new holograms requires negligible computation time (<5 ms on CPU, $\mu$s scale on GPU).

Significance and Claims
The paper positions this work as a significant advancement toward the practical deployment of full-field holography, particularly for augmented and mixed reality. The authors claim that by embedding interference engineering directly into the CGH computation, they bridge the gap between laboratory holography and compact display integration.

Key claims regarding significance include:

• Removal of a Long-Standing Barrier: The method addresses the ZOD problem without the trade-offs of existing solutions (bulky optics, reduced FOV, or efficiency loss).
• Enabling High-Fidelity Engines: By facilitating the removal of the central bright spot, the technique supports the development of compact, high-fidelity holographic engines for near-eye displays.
• Scalability: The approach is adaptable to various phase modulators (provided fill factor >50%) and can be extended to full-color holography via time-multiplexing and wavelength-specific corrections.
• Conceptual Shift: The work represents a shift from external optical filtering to internal phase-engineering for zero-order suppression.

The authors note that residual ZOD after compensation is attributed to physical limitations such as laser polarization purity, SLM fill factor, and pixel electronics flickering, rather than flaws in the algorithm itself. They conclude that this method offers the best available compromise for suppressing zero-order artifacts in phase-only CGH.