Source-linked AI summary
Computational Ghost Imaging
Jeffrey H Shapiro
TL;DR
Ghost imaging traditionally correlates a high-resolution reference measurement with an object-interacting bucket measurement, but this paper develops a single-pixel computational alternative. The approach produces background-free narrowband images, supports 3D sectioning, and indicates that ghost-image formation is classical.
Problem
Ghost imaging conventionally uses both a high spatial-resolution detector and a bucket detector, motivating whether object information can instead be recovered using only a single-pixel detector.
Method
The method deterministically modulates a cw laser, computes the reference intensity-fluctuation pattern by diffraction theory, and correlates it with the bucket-detector photocurrent.
Results
The computational ghost imager yields background-free images, controllable resolution and field of view, and 3D sectioning using precomputed patterns for multiple propagation distances.
Takeaways & Limitations
Using one light beam and one detector rules out interpreting the computational ghost image as arising from nonlocal two-photon interference, underscoring its classical nature.
Takeaways & Limitations
The analysis assumes the narrowband regime and a sufficiently small pinhole approximation for the detector measurement.
Abstract
from arXiv · showhide
Ghost-imaging experiments correlate the outputs from two photodetectors: a high spatial-resolution (scanning pinhole or CCD camera) detector that measures a field which has not interacted with the object to be imaged, and a bucket (single-pixel) detector that collects a field that has interacted with the object. We describe a computational ghost-imaging arrangement that uses only a single-pixel detector. This configuration affords background-free imagery in the narrowband limit and a 3D sectioning capability. It clearly indicates the classical nature of ghost-image formation.