Deep ghost phase imaging

Deep-learning-based single-pixel phase imaging is proposed. The method, termed deep ghost phase imaging (DGPI), succeeds the advantages of computational ghost imaging, i.e., has the phase imaging quality with high signal-to-noise ratio derived from the Fellgett's multiplex advantage and the point-like detection of diffracted light from objects. A deep convolutional neural network is learned to output a desired phase distribution from an input of a defocused intensity distribution reconstructed by the single-pixel imaging theory. Compared to the conventional interferometric and transport-of-intensity approaches to single-pixel phase imaging, the DGPI requires neither additional intensity measurements nor explicit approximations. The effects of defocus distance and light level are investigated by numerical simulation and an optical experiment confirms the feasibility of the DGPI.

Single-shot lensless imaging with fresnel zone aperture and incoherent illumination

Lensless imaging eliminates the need for geometric isomorphism between a scene and an image while allowing the construction of compact, lightweight imaging systems. However, a challenging inverse problem remains due to the low reconstructed signal-to-noise ratio. Current implementations require multiple masks or multiple shots to denoise the reconstruction. We propose single-shot lensless imaging with a Fresnel zone aperture and incoherent illumination. By using the Fresnel zone aperture to encode the incoherent rays in wavefront-like form, the captured pattern has the same form as the inline hologram. Since conventional backpropagation reconstruction is troubled by the twin-image problem, we show that the compressive sensing algorithm is effective in removing this twin-image artifact due to the sparsity in natural scenes. The reconstruction with a significantly improved signal-to-noise ratio from a single-shot image promotes a camera architecture that is flat and reliable in its structure and free of the need for strict calibration.

Learning to synthesize: robust phase retrieval at low photon counts

The quality of inverse problem solutions obtained through deep learning is limited by the nature of the priors learned from examples presented during the training phase. Particularly in the case of quantitative phase retrieval, spatial frequencies that are underrepresented in the training database, most often at the high band, tend to be suppressed in the reconstruction. Ad hoc solutions have been proposed, such as pre-amplifying the high spatial frequencies in the examples; however, while that strategy improves the resolution, it also leads to high-frequency artefacts, as well as low-frequency distortions in the reconstructions. Here, we present a new approach that learns separately how to handle the two frequency bands, low and high, and learns how to synthesize these two bands into full-band reconstructions. We show that this "learning to synthesize" (LS) method yields phase reconstructions of high spatial resolution and without artefacts and that it is resilient to high-noise conditions, e.g., in the case of very low photon flux. In addition to the problem of quantitative phase retrieval, the LS method is applicable, in principle, to any inverse problem where the forward operator treats different frequency bands unevenly, i.e., is ill-posed.

Probing shallower: perceptual loss trained Phase Extraction Neural Network (PLT-PhENN) for artifact-free reconstruction at low photon budget

Deep neural networks (DNNs) are efficient solvers for ill-posed problems and have been shown to outperform classical optimization techniques in several computational imaging problems. In supervised mode, DNNs are trained by minimizing a measure of the difference between their actual output and their desired output; the choice of measure, referred to as "loss function," severely impacts performance and generalization ability. In a recent paper [A. Goy et al., Phys. Rev. Lett. 121(24), 243902 (2018)], we showed that DNNs trained with the negative Pearson correlation coefficient (NPCC) as the loss function are particularly fit for photon-starved phase-retrieval problems, though the reconstructions are manifestly deficient at high spatial frequencies. In this paper, we show that reconstructions by DNNs trained with default feature loss (defined at VGG layer ReLU-22) contain more fine details; however, grid-like artifacts appear and are enhanced as photon counts become very low. Two additional key findings related to these artifacts are presented here. First, the frequency signature of the artifacts depends on the VGG's inner layer that perceptual loss is defined upon, halving with each MaxPooling2D layer deeper in the VGG. Second, VGG ReLU-12 outperforms all other layers as the defining layer for the perceptual loss.

Bi-phase emulsion droplets as dynamic fluid optical systems

Micro-scale optical components play a critical role in many applications, in particular when these components are capable of dynamically responding to different stimuli with a controlled variation of their optical behavior. Here, we discuss the potential of micro-scale bi-phase emulsion droplets as a material platform for dynamic fluid optical components. Such droplets act as liquid compound micro-lenses with dynamically tunable focal lengths. They can be reconfigured to focus or scatter light and form images. In addition, we discuss how these droplets can be used to create iridescent structural color with large angular spectral separation. Experimental demonstrations of the emulsion droplet optics are complemented by theoretical analysis and wave-optical modelling. Finally, we provide evidence of the droplets utility as fluidic optical elements in potential application scenarios.

Low Photon Count Phase Retrieval Using Deep Learning

Imaging systems' performance at low light intensity is affected by shot noise, which becomes increasingly strong as the power of the light source decreases. In this Letter, we experimentally demonstrate the use of deep neural networks to recover objects illuminated with weak light and demonstrate better performance than with the classical Gerchberg-Saxton phase retrieval algorithm for equivalent signal over noise ratio. The prior contained in the training image set can be leveraged by the deep neural network to detect features with a signal over noise ratio close to one. We apply this principle to a phase retrieval problem and show successful recovery of the object's most salient features with as little as one photon per detector pixel on average in the illumination beam. We also show that the phase reconstruction is significantly improved by training the neural network with an initial estimate of the object, as opposed to training it with the raw intensity measurement.

Simultaneous measurement and reconstruction tailoring for quantitative phase imaging

We propose simultaneous measurement and reconstruction tailoring (SMaRT) for quantitative phase imaging; it is a joint optimization approach to inverse problems wherein minimizing the expected end-to-end error yields optimal design parameters for both the measurement and reconstruction processes. Using simulated and experimentally-collected data for a specific scenario, we demonstrate that optimizing the design of the two processes together reduces phase reconstruction error over past techniques that consider these two design problems separately. Our results suggest at times surprising design principles, and our approach can potentially inspire improved solution methods for other inverse problems in optics as well as the natural sciences.

Spectral pre-modulation of training examples enhances the spatial resolution of the phase extraction neural network (PhENN)

The phase extraction neural network (PhENN) [Optica 4, 1117 (2017)] is a computational architecture, based on deep machine learning, for lens-less quantitative phase retrieval from raw intensity data. PhENN is a deep convolutional neural network trained through examples consisting of pairs of true phase objects and their corresponding intensity diffraction patterns; thereafter, given a test raw intensity pattern, PhENN is capable of reconstructing the original phase object robustly, in many cases even for objects outside the database where the training examples were drawn from. Here, we show that the spatial frequency content of the training examples is an important factor limiting PhENN's spatial frequency response. For example, if the training database is relatively sparse in high spatial frequencies, as most natural scenes are, PhENN's ability to resolve fine spatial features in test patterns will be correspondingly limited. To combat this issue, we propose "flattening" the power spectral density of the training examples before presenting them to PhENN. For phase objects following the statistics of natural scenes, we demonstrate experimentally that the spectral pre-modulation method enhances the spatial resolution of PhENN by a factor of 2.

Apparent coherence loss in phase space tomography

A sensor pixel integrates optical intensity across its extent, and we explore the role that this integration plays in phase space tomography. The literature is inconsistent in its treatment of this integration-some approaches model this integration explicitly, some approaches are ambiguous about whether this integration is taken into account, and still some approaches assume pixel values to be point samples of the optical intensity. We show that making a point-sample assumption results in apodization of and thus systematic error in the recovered ambiguity function, leading to underestimating the overall degree of coherence. We explore the severity of this effect using a Gaussian Schell-model source and discuss when this effect, as opposed to noise, is the dominant source of error in the retrieved state of coherence.

Computational integral field spectroscopy with diverse imaging

Integral field spectroscopy (IFS) is a well-established method for measuring spectral intensity data of the form s(x,y,λ), where x, y are spatial coordinates and λ is the wavelength. In most flavors of IFS, there is a trade-off between sampling (x,y) and the measured wavelength band Δλ. Here we present the first, to our knowledge, attempt to overcome this trade-off by use of computational imaging and measurement diversity. We implement diversity by including a grating in our design, which allows rotation of the dispersed spectra between measurements. The raw intensity data captured from the rotated grating positions are then processed by an inverse algorithm that utilizes sparsity in the data. We present simulated results from spatial-spectral data in the experimental dataset. We used non-overlapping portions of the dataset to train our sparsity priors in the form of the dictionary, and to test the reconstruction quality. We found that, depending on the level of noise in the measurement, diversity up to a maximum number of measurements is beneficial in terms of reducing error, and yields diminishing returns if even more measurements are taken.

Fano resonances from coupled whispering-gallery modes in photonic molecules

We present a rigorous investigation of resonant coupling between microspheres based on multipole expansions. The microspheres have diameters in the range of several micrometers and can be used to realize various photonic molecule configurations. We reveal and quantify the interactions between the whispering gallery modes inside individual microspheres and the propagation modes of the entire photonic molecule structures. We show that Fano-like resonances in photonic molecules can be engineered by tuning the coupling between the resonant and radiative modes when the structures are illuminated with simple dipole radiation.

Superoleophilic Titania Nanoparticle Coatings with Fast Fingerprint Decomposition and High Transparency

Low surface tension sebaceous liquids such as human fingerprint oils are readily deposited on high energy surfaces such as clean glass, leaving smudges that significantly lower transparency. There have been several attempts to prevent formation of these dactylograms on glass by employing oil-repellent textured surfaces. However, nanotextured superoleophobic coatings typically scatter visible light, and the intrinsic thermodynamic metastability of the composite superoleophobic state can result in failure of the oil repellency under moderate contact pressure. We develop titania-based porous nanoparticle coatings that are superoleophilic and highly transparent and which exhibit short time scales for decomposition of fingerprint oils under ultraviolet light. The mechanism by which a typical dactylogram is consumed combines wicking of the sebum into the nanoporous titania structure followed by photocatalytic degradation. We envision a wide range of applications because these TiO₂ nanostructured surfaces remain photocatalytically active against fingerprint oils in natural sunlight and are also compatible with flexible glass substrates.

Denoised Wigner distribution deconvolution via low-rank matrix completion

Wigner distribution deconvolution (WDD) is a decades-old method for recovering phase from intensity measurements. Although the technique offers an elegant linear solution to the quadratic phase retrieval problem, it has seen limited adoption due to its high computational/memory requirements and the fact that the technique often exhibits high noise sensitivity. Here, we propose a method for noise suppression in WDD via low-rank noisy matrix completion. Our technique exploits the redundancy of an object's phase space to denoise its WDD reconstruction. We show in model calculations that our technique outperforms other WDD algorithms as well as modern iterative methods for phase retrieval such as ptychography. Our results suggest that a class of phase retrieval techniques relying on regularized direct inversion of ptychographic datasets (instead of iterative reconstruction techniques) can provide accurate quantitative phase information in the presence of high levels of noise.

Digital micromirror device-based laser-illumination Fourier ptychographic microscopy

We report a novel approach to Fourier ptychographic microscopy (FPM) by using a digital micromirror device (DMD) and a coherent laser source (532 nm) for generating spatially modulated sample illumination. Previously demonstrated FPM systems are all based on partially-coherent illumination, which offers limited throughput due to insufficient brightness. Our FPM employs a high power coherent laser source to enable shot-noise limited high-speed imaging. For the first time, a digital micromirror device (DMD), imaged onto the back focal plane of the illumination objective, is used to generate spatially modulated sample illumination field for ptychography. By coding the on/off states of the micromirrors, the illumination plane wave angle can be varied at speeds more than 4 kHz. A set of intensity images, resulting from different oblique illuminations, are used to numerically reconstruct one high-resolution image without obvious laser speckle. Experiments were conducted using a USAF resolution target and a fiber sample, demonstrating high-resolution imaging capability of our system. We envision that our approach, if combined with a coded-aperture compressive-sensing algorithm, will further improve the imaging speed in DMD-based FPM systems.

Conical photonic crystals for enhancing light extraction efficiency from high refractive index materials

We propose, analyze and optimize a two-dimensional conical photonic crystal geometry to enhance light extraction from a high refractive index material, such as an inorganic scintillator. The conical geometry suppresses Fresnel reflections at an optical interface due to adiabatic impedance matching from a gradient index effect. The periodic array of cone structures with a pitch larger than the wavelength of light diffracts light into higher-order modes with different propagating angles, enabling certain photons to overcome total internal reflection (TIR). The numerical simulation shows simultaneous light yield gains relative to a flat surface both below and above the critical angle and how key parameters affect the light extraction efficiency. Our optimized design provides a 46% gain in light yield when the conical photonic crystals are coated on an LSO (cerium-doped lutetium oxyorthosilicate) scintillator.

Focusing and imaging in microsphere-based microscopy

Microsphere-based microscopy systems have garnered lots of recent interest, mainly due to their capacity in focusing light and imaging beyond the diffraction limit. In this paper, we present theoretical foundations for studying the optical performance of such systems by developing a complete theoretical model encompassing the aspects of illumination, sample interaction and imaging/collection. Using this model, we show that surface waves play a significant role in focusing and imaging with the microsphere. We also show that by designing a radially polarized convergent beam, we can focus to a spot smaller than the diffraction limit. By exploiting surface waves, we are able to resolve two dipoles spaced 98 nm apart in simulation using light at a wavelength of 402.292 nm. Using our model, we also explore the effect of beam geometry and polarization on optical resolution and focal spot size, showing that both geometry and polarization greatly affect the shape of the spot.

Polarization-independent Talbot effect

We report the first observation of polarization-independent Talbot effect with a high-density grating for TE and TM polarizations, which is attributed to the identical phases and diffraction efficiencies of the diffraction orders for both polarizations. We introduce the simplified modal method that provides an insightful physical description for explanation of the diffraction efficiency and phase of the polarization-independent Talbot effect. Only two even grating modes can be excited, which determines the diffraction properties of the near-field image. We expect that this theoretical work will be helpful for the tremendous potential applications of the Talbot effect.

Wigner analysis of three dimensional pupil with finite lateral aperture

A three dimensional (3D) pupil is an optical element, most commonly implemented on a volume hologram, that processes the incident optical field on a 3D fashion. Here we analyze the diffraction properties of a 3D pupil with finite lateral aperture in the 4-f imaging system configuration, using the Wigner Distribution Function (WDF) formulation. Since 3D imaging pupil is finite in both lateral and longitudinal directions, the WDF of the volume holographic 4-f imager theoretically predicts distinct Bragg diffraction patterns in phase space. These result in asymmetric profiles of diffracted coherent point spread function between degenerate diffraction and Bragg diffraction, elucidating the fundamental performance of volume holographic imaging. Experimental measurements are also presented, confirming the theoretical predictions.

Empirical concentration bounds for compressive holographic bubble imaging based on a Mie scattering model

We use compressive in-line holography to image air bubbles in water and investigate the effect of bubble concentration on reconstruction performance by simulation. Our forward model treats bubbles as finite spheres and uses Mie scattering to compute the scattered field in a physically rigorous manner. Although no simple analytical bounds on maximum concentration can be derived within the classical compressed sensing framework due to the complexity of the forward model, the receiver operating characteristic (ROC) curves in our simulation provide an empirical concentration bound for accurate bubble detection by compressive holography at different noise levels, resulting in a maximum tolerable concentration much higher than the traditional back-propagation method.

Temporal reshaping of two-dimensional pulses

An analytic study of complete cylindrical focusing of pulses in two dimensions is presented, and compared with the analogous three-dimensional case of focusing over a complete sphere. Such behavior is relevant for understanding the limiting performance of ultrafast, planar photonic and plasmonic devices. A particular spectral distribution is assumed that contains finite energy. Separate ingoing and outgoing pulsed waves are considered, along with the combination that would be generated in free space by an ingoing wave. It is shown that for the two dimensional case, in order to produce a temporally symmetrical pulse at the focus, an asymmetric pulse must be launched. A symmetrical outgoing pulse is generated from a source with asymmetric time behavior, or an anti-symmetric input pulse. These results are very different from the corresponding three-dimensional case, and imply fundamental limitations on the performance of ultrafast, tightly focused, two-dimensional devices.

Background suppression in long-distance imaging using volume hologram filters

We performed experiments using a volume hologram filter (VHF) coupled with a telephoto objective lens to detect weak distant signals masked by strong background noise. The VHF was able to selectively pass light originating from a certain distance while attenuating background noise contributions from other distances, resulting in a higher signal-to-noise ratio (SNR). The proposed method is useful in remote sensing applications such as daytime artificial satellite and space debris detection.

Phase imaging for absorptive phase objects using hybrid uniform and structured illumination transport of intensity equation

Transport of intensity equation (TIE) has been a popular and convenient phase imaging method that retrieves phase profile from the measurement of intensity differentials. Conventional 2-shot uniform illumination TIE can give reliable inversion of the phase from intensity in many situations of practical interest; however, it has a null space consisting of fields with non-zero circulation of the Poynting vector. Here, we propose the hybrid illumination TIE method to disambiguate such objects. By comparing the diffraction signals using uniform and structured (sinusoidal) illumination patterns, we obtain a modulation-induced signal that depends solely on the phase gradient. In this way, we also increase signal sensitivity in the low spatial frequency region.

Low-noise phase imaging by hybrid uniform and structured illumination transport of intensity equation

We demonstrate a new approach to the transport of intensity equation (TIE) phase retrieval method which uses structured illumination to improve low-frequency noise performance. In the hybrid scheme, two phase images are acquired: one with uniform and one with sinusoidal grating illumination intensity. The former preserves the high spatial frequency features of the phase best, whereas the latter dramatically increase the response at low spatial frequencies (where traditional TIE notoriously suffers). We then theoretically prove the design of a spectral filter that optimally combines the two phase results while suppressing noise. The combination of uniformly and structured illuminated TIE (hybrid TIE) phase imaging is experimentally demonstrated optically with a calibrated pure phase object.

Phase imaging using shifted wavefront sensor images

We propose a new approach to the complete retrieval of a coherent field (amplitude and phase) using the same hardware configuration as a Shack-Hartmann sensor but with two modifications: first, we add a transversally shifted measurement to resolve ambiguities in the measured phase; and second, we employ factored form descent (FFD), an inverse algorithm for coherence retrieval, with a hard rank constraint. We verified the proposed approach using both numerical simulations and experiments.