All-Optical Naked-Eye Ghost Imaging

Gao Wang, Huaibin Zheng, Zhiguo Tang, Yu Zhou, Hui Chen, Jianbin Liu, Yuchen He, Yuan Yuan, Fuli Li, Zhuo Xu

TL;DR

Researchers have developed a "naked-eye" ghost imaging system that reconstructs images of moving, colored objects without any electronic computers or optoelectronic detectors. By using a specially designed rotating modulation disk and leveraging the human eye's natural persistence of vision, the system performs complex correlations entirely through light-light interactions.

What Problem Does This Paper Address?

Ghost imaging (GI) is a technique where an image is reconstructed by correlating two beams of light: one that interacts with an object but has no spatial resolution (the "bucket" signal), and one that never sees the object but has a known spatial pattern (the "reference").

Traditionally, this process is data-heavy and slow. It requires high-speed photodetectors, a spatial light modulator (or a CCD camera), and a computer to crunch the numbers and perform the cross-correlation. Previous attempts at "naked-eye" ghost imaging—where the eye replaces the computer—suffered from extremely low contrast because the reconstructed image was often "washed out" by the reference light. This paper aims to remove the electronic bottleneck entirely while achieving high-contrast, real-time imaging.

Methods

The authors propose an all-optical correlation scheme. Instead of a computer multiplying signal values by pattern values, the "calculation" happens physically:

  1. Modulation: A light source (like a blue laser or white LED) passes through a rotating "light-modulation disk" containing specific patterns.
  2. Interaction: This patterned light hits the object. The transmitted light is then collected by a "bucket" collector (like a fiber or ground glass) which destroys the spatial information but keeps the total intensity.
  3. Correlation: This collected light is sent back to pass through the same (or an identical) rotating disk. Because the disk is rotating, the light passing through it at any moment is automatically synchronized with the pattern that originally illuminated the object.
  4. Integration: The human eye (or a CCD camera mimicking the eye) integrates these light pulses over a short window of time (about 0.2 seconds). This "persistence of vision" effectively performs the summation part of the ghost imaging algorithm.

To solve the contrast problem, the team redesigned the mathematical basis of their patterns. They used a reduced Hadamard matrix, abandoning the first row and column of a standard Hadamard matrix to eliminate the "background" light that usually blurs the results. They also implemented a spatial partitioning strategy, dividing the image into rows and cells to boost local contrast and resolution.

Datasets

As this is an experimental physics paper rather than a machine learning study, standard datasets were not used. Instead, the authors evaluated their system using:

  • Static and Moving Objects: Cutouts of the letters "X", "J", "T", and "U".
  • Resolution: The objects were defined at a resolution of $35 \times 35$ pixels.
  • Color: For the advanced setup, they used colored versions of these letters (Red "X", Green "J", Blue "T", and White "U") to demonstrate multi-wavelength capabilities.

Key Insights and Ideas

  • The Eye as a Processor: The most elegant insight is treating the human retina as the integration component of a computational imaging system. By matching the disk's rotation speed to the eye's persistence time, the "math" of ghost imaging happens in the viewer's brain.
  • Ground Glass as a "Bucket": In their color setup, the researchers used ground glass to scatter light. This acts as an "optical homogenizer," serving the role of a bucket detector without needing complex fiber optics or electronic sensors.
  • Hardware-Level Optimization: By physically removing the "DC component" (the first row/column of the Hadamard matrix) from the physical disk, they moved the contrast-enhancement step from post-processing software directly into the optical hardware.

Key Results

  • Real-Time Imaging: The system successfully produced high-contrast images of moving objects that could be seen directly by the naked eye.
  • Color Support: Unlike many ghost imaging systems that are monochromatic, the second iteration of their setup (using white LEDs and ground glass) successfully reconstructed multi-colored objects.
  • Speed and Resolution: The system achieved $35 \times 35$ pixel resolution. By using the optimized Hadamard scanning, they significantly boosted the imaging speed compared to traditional random-speckle ghost imaging.
  • Persistence Settings: For the CCD camera used to mimic the eye, an exposure time of 0.2 seconds was found to be the "sweet spot" for mimicking human night vision and producing a stable image.

Limitations and Future Work

  • Efficiency: The authors noted that the initial setup using optical fiber coupling was inefficient and difficult to align. While the ground glass setup improved this, light throughput remains a challenge for all-optical systems.
  • Resolution Scaling: While $35 \times 35$ is a proof of concept, scaling to high-definition resolutions would require much larger disks or faster modulation speeds.
  • Broad Application: The authors suggest this technique isn't limited to visible light; it could potentially be applied to X-rays, atoms, or electron imaging, where electronic detectors are often expensive or difficult to implement.

Summary updated

Abstract

Ghost imaging is usually based on the optoelectronic process and electronic computing. A new ghost imaging approach is put forward in the paper that avoids any optoelectronic or electronic process. Instead, the proposed scheme exploits all-optical correlation and the vision persistence effect to generate images observed by naked eyes. To realize high contrast naked-eye ghost imaging, a special pattern-scanning architecture on a low-speed light-modulation disk is designed, which also enables high-resolution imaging with lower-order Hadamard vectors and boosts the imaging speed. With this approach, we realize high-contrast real-time naked-eye ghost imaging for moving colored objects.
BibTeX
@inproceedings{wang2020all,
  author = {Gao Wang and Huaibin Zheng and Zhiguo Tang and Yu Zhou and Hui Chen and Jianbin Liu and Yuchen He and Yuan Yuan and Fuli Li and Zhuo Xu},
  title = {{All-Optical Naked-Eye Ghost Imaging}},
  booktitle = {Scientific},
  year = {2020},
  doi = {10.1038/s41598-020-59263-1},
}

Loading related content...

Figure