[Paper Review] Computational ghost imaging versus imaging laser radar for 3D imaging

Nicholas D. Hardy, Jeffrey H. Shapiro|arXiv (Cornell University)|Dec 13, 2012

Random lasers and scattering media23 references113 citations

TL;DR

This paper compares a pulsed computational ghost imaging system with a pulsed floodlight-illumination imaging laser radar for 3D remote sensing of rough-surfaced targets under atmospheric turbulence. Using analytical models, it shows that both systems achieve comparable spatial resolution and saturation signal-to-noise ratio (SNR) when matched in aperture and pulse energy, but the laser radar significantly outperforms ghost imaging in shot-noise-limited SNR by a factor equal to the number of spatial resolution cells on the target. The key advantage of ghost imaging emerges in bistatic configurations where the transmitter is in weak turbulence and the receiver in strong turbulence, or when using single-pixel detectors in multi-static networks.

ABSTRACT

Ghost imaging has been receiving increasing interest for possible use as a remote-sensing system. There has been little comparison, however, between ghost imaging and the imaging laser radars with which it would be competing. Toward that end, this paper presents a performance comparison between a pulsed, computational ghost imager and a pulsed, floodlight-illumination imaging laser radar. Both are considered for range-resolving (3D) imaging of a collection of rough-surfaced objects at standoff ranges in the presence of atmospheric turbulence. Their spatial resolutions and signal-to-noise ratios are evaluated as functions of the system parameters, and these results are used to assess each system's performance trade-offs. Scenarios in which a reflective ghost-imaging system has advantages over a laser radar are identified.

Motivation & Objective

• To evaluate and compare the performance of computational ghost imaging and imaging laser radar for 3D remote sensing.
• To analyze spatial resolution and signal-to-noise ratio (SNR) under atmospheric turbulence for both systems.
• To identify scenarios where ghost imaging offers advantages over laser radar, particularly in bistatic configurations or with limited detector technology.
• To assess the impact of detector and aperture size on resolution and SNR trade-offs.
• To determine whether compressed sensing can mitigate the SNR disadvantage of correlation-based ghost imaging.

Proposed method

• Uses a pulsed, computational ghost imaging setup with a spatial light modulator (SLM) to generate structured illumination and a single-pixel bucket detector for intensity collection.
• Employs diffraction theory to compute the reference intensity pattern for each pulse, eliminating the need for a physical reference beam.
• Models the laser radar using a floodlight-illumination approach with a focusing lens and CCD array to capture 3D images via time-gated detection.
• Applies statistical optics and speckle statistics to model target return under fully developed laser speckle and atmospheric turbulence.
• Derives analytical expressions for spatial resolution and SNR, accounting for turbulence in both source-to-target and target-to-receiver paths.
• Performs a fair comparison by matching average photon number per pulse (NT) and number of pulses (N) between systems.

Experimental results

Research questions

• RQ1 How do computational ghost imaging and imaging laser radar compare in spatial resolution under atmospheric turbulence?
• RQ2 What is the signal-to-noise ratio (SNR) performance of each system under shot-noise-limited and saturated conditions?
• RQ3 In what scenarios does ghost imaging outperform laser radar, particularly regarding turbulence distribution and system configuration?
• RQ4 How does the number of spatial resolution cells affect the SNR advantage of laser radar over ghost imaging in the shot-noise-limited regime?
• RQ5 Can compressed sensing reduce the number of pulses required for ghost imaging, thereby mitigating its SNR disadvantage?

Key findings

• Both systems achieve equal spatial resolution when matched in aperture and pulse energy, with resolution limited by turbulence in the source-to-target path for ghost imaging and target-to-receiver path for laser radar.
• In the shot-noise-limited regime, the laser radar outperforms ghost imaging by a factor approximately equal to the number of spatial resolution cells on the target.
• At high pulse energy and number of pulses, both systems saturate at the same SNR, indicating no long-term advantage for either under saturation.
• Ghost imaging has infinite depth of focus, while laser radar suffers from defocus when k0r2ℓ/2z ≫ 1, limiting its range performance.
• The primary advantage of ghost imaging arises in bistatic configurations where the transmitter is in weak turbulence and the receiver in strong turbulence, or in multi-static networks with single-pixel detectors.
• Compressed sensing techniques could reduce the number of pulses needed for ghost imaging, potentially mitigating its SNR disadvantage in practical applications.