Imaging With Nature: Compressive Imaging Using a Multiply Scattering Medium

Abstract: The recent theory of compressive sensing leverages upon the structure of signals to acquire them with much fewer measurements than was previously thought necessary, and certainly well below the traditional Nyquist-Shannon sampling rate. However, most implementations developed to take advantage of this framework revolve around controlling the measurements with carefully engineered material or acquisition sequences. Instead, we use the natural randomness of wave propagation through multiply scattering media as an optimal and instantaneous compressive imaging mechanism. Waves reflected from an object are detected after propagation through a well-characterized complex medium. Each local measurement thus contains global information about the object, yielding a purely analog compressive sensing method. We experimentally demonstrate the effectiveness of the proposed approach for optical imaging by using a 300-micrometer thick layer of white paint as the compressive imaging device. Scattering media are thus promising candidates for designing efficient and compact compressive imagers.

• The paper demonstrates how compressive sensing exploits the randomness of a ZnO scattering medium to overcome traditional Nyquist constraints.
• It calibrates the system's transmission matrix using an SLM to multiplex signals in parallel for efficient sparse recovery.
• Experimental results show a phase transition in reconstruction success, mirroring the behavior of ideal Gaussian random matrices.

Compressive Imaging Using Multiply Scattering Media

The paper entitled "Imaging With Nature: Compressive Imaging Using a Multiply Scattering Medium" presents a novel approach to optimize compressive sensing (CS) by leveraging natural materials as analog randomizers. The method employs the intrinsic randomness of wave propagation through multiply scattering media to achieve efficient compressive imaging, specifically in the domain of optical imaging.

Key Contributions

The research centers on the application of CS — a methodology that circumvents traditional Nyquist-Shannon sampling constraints by exploiting the sparse nature of data. This study uniquely utilizes the inherent randomness posed by multiply scattering media to perform parallel multiplexing of signals. The paper proposes the use of a layer of Zinc Oxide (ZnO), akin to white paint, as a scattering medium, offering a compelling alternative to engineered randomization devices like digital micromirror devices (DMD) and spatial light modulators (SLM).

Experimental Setup and Results

The experimental setup involves a Spatial Light Modulator to initiate wavefronts, which are then scattered through a ZnO medium, generating a speckle pattern. This pattern acts as an input-output map, enabling sparse signal reconstruction through calibration of the system's transmission matrix (TM).

Numerical findings indicate successful reconstruction rates for sparse signals significantly below the conventional Nyquist rate. This system accomplishes compressive sensing with a reduced number of measurements, achieving competitive sampling rates. The results demonstrated strong efficacy, with a visible phase transition in reconstruction success as a function of sensor density and sparsity level. The CS framework, specifically the TM measured at the calibration phase, exhibited a close resemblance to an ideal Gaussian random matrix, underscoring its potential in sparse signal recovery.

Implications and Future Directions

This methodology holds promise for advancing compact optical imaging devices and can be transposed across other frequency domains like THz and RF or ultrasound imaging. Notably, the proposed approach facilitates parallel acquisition, substantially decreasing acquisition times compared to sequential CS systems, such as single-pixel cameras.

The practical application requires a calibration phase for TM estimation before conducting compressive imaging. However, once calibrated, the system benefits from significant implementation ease compared to other engineered devices.

Future research may consider refining the calibration technique and reconstruction algorithms to improve performance across varying scattering media and refine the robustness of the system in high noise environments. Moreover, exploring compressive phase retrieval techniques could further streamline the process by minimizing intensity measurements required during acquisition.

In conclusion, this paper suggests a promising direction for compressive sensing applications by merging the stochastic aspects of natural scattering media with advanced mathematical reconstructive theories. Its adaptability and efficiency could significantly impact various sensor-limited imaging systems.