🔬 optics

Gabor Holography Reinvented

This paper presents an optical solution to the long-standing twin-image problem in Gabor holography, effectively reinventing the on-axis method to eliminate the need for iterative phase retrieval or machine learning approaches while preserving its inherent robustness and practical advantages.

Original authors: Jesper Glückstad

Published 2026-02-24

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Jesper Glückstad

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Problem: The "Ghost" in the Machine

Imagine you are trying to take a photograph of a beautiful, transparent glass sculpture. But there's a catch: every time you take a picture, a faint, blurry "ghost" of the sculpture appears right on top of the real one, making the image look messy and confusing.

This is exactly the problem with Gabor Holography, a method invented by Nobel Prize winner Dennis Gabor in the 1940s.

How it works: It uses a single beam of light to both illuminate an object and act as a reference. It's like shining a flashlight on a statue while looking at the reflection in a mirror to figure out the statue's shape.
The Flaw: Because the light travels in a straight line (on-axis), the "real" image and a "twin" (ghost) image overlap perfectly. It's like trying to read a book while someone is holding a second, blurry copy of the same book directly over your eyes.

For decades, scientists tried to fix this using complex math (iterative algorithms) or AI (machine learning). But these methods are slow, require massive computing power, or need thousands of training photos. Dennis Gabor himself eventually gave up on this method because the "ghost" was too hard to remove.

The Solution: A "Traffic Cop" for Light

The author, Jesper Glückstad, has "reinvented" Gabor Holography. Instead of using math to clean up the mess later, he uses optics (physical lenses and filters) to separate the real image from the ghost before the camera even takes the picture.

Here is the analogy:

Imagine the light carrying the image of your object is a crowd of people walking down a hallway.

The Real Image is the main group of people walking straight ahead.
The Twin Image (Ghost) is a group of people walking right next to them, overlapping and causing a traffic jam.
The Old Way: You take a photo of the jam, then use a computer program to try to digitally erase the extra people. It's slow and often leaves smudges.
The New Way (The Reinvention): You install a special curved wall (an Axicon) in the hallway. This wall gently pushes the "Ghost" group of people to the side, into a separate lane, while the "Real" group keeps walking straight.

Now, when you take the photo, the ghost is in a different lane and doesn't block the view at all. The image is instantly clear.

How It Works (The "Magic" Lens)

The paper introduces a specific tool called an Axicon. Think of this as a special cone-shaped lens (or a ring-shaped filter) placed in the path of the light.

The Trick: This lens acts like a prism that bends light based on its "spin" or angle.
The Result: It takes the "Ghost" image (which is mathematically the opposite of the real image) and physically moves it in a circle around the center.
The Cleanup: The camera sensor sees the real object in the center. The ghost is now floating in a ring around the edge. The computer simply ignores the ring and keeps the center.

Because the ghost is moved by physics (the lens) rather than math (software), the process is:

Instant: No waiting for slow computer calculations.
Robust: It works even if the equipment shakes a little.
Simple: It doesn't need massive AI training data.

Why This Matters

The paper shows two things:

Computer Simulations: They simulated a Danish coin (with one side as the shape and the other as the texture). The old method showed a blurry mess; the new method showed a crisp, perfect coin.
Real-World Test: They put a water droplet under a microscope. The new method successfully removed the ghost image, revealing the clear shape and phase of the droplet.

The Bottom Line

This paper solves a 70-year-old problem in holography. Dennis Gabor invented the technique but abandoned it because of the "twin-image" ghost. This new "Reinvented" version uses a clever optical trick (a cone-shaped lens) to physically push the ghost away, leaving us with a crystal-clear hologram without needing super-computers or complex AI.

It's like finally finding a way to turn off the "ghosting" on your TV screen just by adjusting the antenna, rather than trying to fix the picture with a computer program.

1. Problem Statement

The paper addresses the twin-image problem, a fundamental limitation inherent to in-line (on-axis) Gabor holography.

Context: While in-line holography offers significant advantages over off-axis holography—such as common-path robustness, relaxed mechanical stability requirements, lower sensor resolution needs, and shorter exposure times—it suffers from the superposition of the real image, the virtual (twin) image, and the zero-order background.
The Challenge: In the standard holographic equation $I = |O+R|^2 = OO^* + RR^* + OR^* + O^*R$ , the term $O^*R$ represents the twin image. Because the object and reference waves propagate along the same axis, these terms overlap spatially in the Fourier domain, making it impossible to filter them out using standard spatial filtering.
Current Limitations: Existing solutions rely on iterative numerical phase retrieval (e.g., Gerchberg-Saxton algorithms) or machine learning (deep neural networks). These methods are computationally expensive, suffer from non-convex optimization issues (risk of local minima), require massive training datasets, or are too slow for real-time applications.

2. Methodology

The author proposes a novel optical solution to eliminate the twin image without relying on iterative computation or machine learning. The core methodology involves the strategic placement of a phase-only transfer function (specifically an axicon) in the spatial frequency domain.

Theoretical Framework:
- The system is modeled using a generalized transfer function $H$ and its conjugate $H^*$ .
- The goal is to design $H$ such that the twin-image term is spatially shifted away from the object reconstruction term in the output plane.
- The paper utilizes the stationary phase method to relate the spatial derivative of the phase to the local spatial frequency, allowing for the engineering of specific transfer functions.
The Axicon Phase Shaper:
- A diffractive axicon is introduced in the spatial frequency domain (Fourier plane).
- The axicon imparts a radial phase profile: $H = \exp(i\phi_0) \exp(-i 2\pi r' / d)$ .
- This creates a convolution kernel $h^*$ that acts as a ring-shaped delta function: $h^* \simeq \exp(-i\phi_0) \delta(r - r_0)$ .
- Mechanism of Separation:
  1. The object information is convolved with the ring kernel, shifting the holographic terms into a circular band of radius $r_0$ .
  2. During reconstruction, the twin-image term ( $O^*$ ) undergoes a double convolution with the kernel (or its conjugate), pushing it to a ring of radius $2r_0$ .
  3. The real object term remains centered or in a distinct band, while the twin image is physically separated into a different radial zone.
Reconstruction Process:
- By digitally or optically discarding the signal components located at radius $2r_0$ (where the twin image resides), the system isolates the object term $O$ .
- The remaining signal contains the object $O$ plus a constant offset and a reduced self-interference term ( $|O|^2$ ), which is significantly less problematic than the overlapping twin image.

3. Key Contributions

First Optical Solution: This is the first method to overcome the twin-image problem in Gabor holography purely by optical means, eliminating the need for iterative numerical phase retrieval or heavy machine learning training.
Generalized Transfer Function Design: The paper demonstrates that any phase shaper capable of generating a parametric kernel that pushes the twin image outside the signal area is valid, though the axicon is used as the primary example.
Phase-Only Object Optimization: The method is shown to be particularly effective for phase-only objects, where the reconstruction yields a perfectly correct phase profile superimposed on a constant offset.
Hybrid Modalities: The approach is applicable across various holographic instantiations, including All-Optical Holography (OH), Computer-Generated Holography (CGH), and Digital Holography (DH).

4. Results

The paper validates the method through both numerical simulations and experimental proof-of-concept:

Numerical Simulation (Figure 5):
- A complex-valued object (encoding two sides of a Danish coin) was simulated.
- Standard Gabor Holography: Showed severe distortion due to the overlapping twin image and self-interference.
- Reinvented Gabor Holography: Successfully reconstructed the amplitude and phase with high accuracy. The twin image was completely separated and removed, leaving only a minor self-interference artifact.
Experimental Verification (Figure 6):
- A water droplet on a microscope slide was imaged using a setup with a refractive axicon phase shaper.
- The sensor read-out clearly showed the characteristic annular band of holographic terms.
- The digital reconstruction of the water droplet's amplitude and phase demonstrated the effective optical removal of the twin image, confirming the theoretical predictions.
Reflective 3D Replay (Figure 7):
- A hardware embodiment was demonstrated in a reflection geometry, showing the feasibility of analog optical recording and reconstruction using the axicon method.

5. Significance

Revival of Gabor Holography: This work effectively "reinvents" the original on-axis holography concept, allowing it to be used in modern applications without the historical drawback of the twin image.
Real-Time Capability: By removing the need for iterative algorithms, the method enables high-speed, real-time holographic imaging, which is critical for dynamic applications like particle trapping, flow cytometry, and live microscopy.
Robustness and Simplicity: The method retains the inherent advantages of in-line holography (common-path stability, relaxed coherence requirements) while adding the clarity of off-axis holography.
Broad Applicability: The technique is versatile, applicable to transparent, semi-opaque, and opaque objects, and can be integrated into existing digital holographic microscopy (DHM) systems with the addition of a phase shaper element.

In conclusion, Jesper Glückstad presents a paradigm shift in holography, solving a 70-year-old problem through optical engineering rather than computational brute force, thereby unlocking the full potential of in-line holography for modern photonics.