Aurora Maccarone

Underwater depth imaging using time-correlated single photon counting

We investigate the potential of a depth imaging system for underwater environments. This system is based on the timeof- flight approach and the time correlated single-photon counting (TCSPC) technique. We report laboratory-based measurements and explore the potential of achieving sub-centimeter xyz resolution at 10’s meters stand-off distances. Initial laboratory-based experiments demonstrate depth imaging performed over distances of up to 1.8 meters and under a variety of scattering conditions. The system comprised a monostatic transceiver unit, a fiber-coupled supercontinuum laser with a wavelength tunable acousto-optic filter, and a fiber-coupled individual silicon single-photon avalanche diode (SPAD). The scanning in xy was performed using a pair of galvonometer mirrors directing both illumination and scattered returns via a coaxial optical configuration. Target objects were placed in a 110 liter capacity tank and depth images were acquired through approximately 1.7 meters of water containing different concentrations of scattering agent. Depth images were acquired in clear and highly scattering water using per-pixel acquisition times in the range 0.5-100 ms at average optical powers in the range 0.8 nW to 120 μW. Based on the laboratory measurements, estimations of potential performance, including maximum range possible, were performed with a model based on the LIDAR equation. These predictions will be presented for different levels of scattering agent concentration, optical powers, wavelengths and comparisons made with naturally occurring environments. The experimental and theoretical results indicate that the TCSPC technique has potential for highresolution underwater depth profile measurements.

Object Depth Profile and Reflectivity Restoration From Sparse Single-Photon Data Acquired in Underwater Environments

This paper presents two new algorithms for the joint restoration of depth and reflectivity (DR) images constructed from time-correlated single-photon counting measurements. Two extreme cases are considered: 1) a reduced acquisition time that leads to very low photon counts; and 2) imaging in a highly attenuating environment (such as a turbid medium), which makes the reflectivity estimation more difficult at increasing range. Adopting a Bayesian approach, the Poisson distributed observations are combined with prior distributions about the parameters of interest, to build the joint posterior distribution. More precisely, two Markov random field (MRF) priors enforcing spatial correlations are assigned to the DR images. Under some justified assumptions, the restoration problem (regularized likelihood) reduces to a convex formulation with respect to each of the parameters of interest. This problem is first solved using an adaptive Markov chain Monte Carlo (MCMC) algorithm that approximates the minimum mean square parameter estimators. This algorithm is fully automatic since it adjusts the parameters of the MRFs by maximum marginal likelihood estimation. However, the MCMC-based algorithm exhibits a relatively long computational time. The second algorithm deals with this issue and is based on a coordinate descent algorithm. Results on single-photon depth data from laboratory-based underwater measurements demonstrate the benefit of the proposed strategy that improves the quality of the estimated DR images.

Robust Spectral Unmixing of Sparse Multispectral Lidar Waveforms Using Gamma Markov Random Fields

This paper presents a new Bayesian spectral unmixing algorithm to analyze remote scenes sensed via sparse multispectral Lidar measurements. To a first approximation, in the presence of a target, each Lidar waveform consists of a main peak, whose position depends on the target distance and whose amplitude depends on the wavelength of the laser source considered (i.e., on the target reflectivity). Besides, these temporal responses are usually assumed to be corrupted by Poisson noise in the low photon count regime. When considering multiple wavelengths, it becomes possible to use spectral information in order to identify and quantify the main materials in the scene, in addition to estimation of the Lidar-based range profiles. Due to its anomaly detection capability, the proposed hierarchical Bayesian model, coupled with an efficient Markov chain Monte Carlo algorithm, allows robust estimation of depth images together with abundance and outlier maps associated with the observed three-dimensional scene. The proposed methodology is illustrated via experiments conducted with real multispectral Lidar data acquired in a controlled environment. The results demonstrate the possibility to unmix spectral responses constructed from extremely sparse photon counts (less than 10 photons per pixel and band).

Joint range estimation and spectral classification for 3D scene reconstruction using multispectral Lidar waveforms

This paper presents a new Bayesian classification method to analyse remote scenes sensed via multispectral Lidar measurements. To a first approximation, each Lidar waveform mainly consists of the temporal signature of the observed target, which depends on the wavelength of the laser source considered and which is corrupted by Poisson noise. By sensing the scene at several wavelengths, we expect a more accurate target range estimation and a more efficient spectral analysis of the scene. Thanks to its spectral classification capability, the proposed hierarchical Bayesian model, coupled with an efficient Markov chain Monte Carlo algorithm, allows the estimation of depth images together with reflectivity-based scene segmentation images. The proposed methodology is illustrated via experiments conducted with real multispectral Lidar data.

Joint spectral clustering and range estimation for 3D scene reconstruction using multispectral lidar waveforms

This paper presents a new Bayesian clustering method to analyse remote scenes sensed via multispectral Lidar measurements. To a first approximation, each Lidar waveform mainly consists of the temporal signature of the observed target, which depends on the wavelength of the laser source considered and which is corrupted by Poisson noise. By sensing the scene at several wavelengths, we expect a more accurate target range estimation and a more efficient spectral analysis of the scene. Thanks to its spectral classification capability, the proposed hierarchical Bayesian model, coupled with an efficient Markov chain Monte Carlo algorithm, allows the estimation of depth images together with reflectivity-based scene segmentation images. The proposed methodology is illustrated via experiments conducted with real multispectral Lidar data.

Depth imaging in highly scattering underwater environments using time-correlated single-photon counting

This paper presents an optical depth imaging system optimized for highly scattering environments such as underwater. The system is based on the time-correlated single-photon counting (TCSPC) technique and the time-of-flight approach. Laboratory-based measurements demonstrate the potential of underwater depth imaging, with specific attention given to environments with a high level of scattering. The optical system comprised a monostatic transceiver unit, a fiber-coupled supercontinuum laser source with a wavelength tunable acousto-optic filter (AOTF), and a fiber-coupled single-element silicon single-photon avalanche diode (SPAD) detector. In the optical system, the transmit and receive channels in the transceiver unit were overlapped in a coaxial optical configuration. The targets were placed in a 1.75 meter long tank, and raster scanned using two galvo-mirrors. Laboratory-based experiments demonstrate depth profiling performed with up to nine attenuation lengths between the transceiver and target. All of the measurements were taken with an average laser power of less than 1mW. Initially, the data was processed using a straightforward pixel-wise cross-correlation of the return timing signal with the system instrumental timing response. More advanced algorithms were then used to process these cross-correlation results. These results illustrate the potential for the reconstruction of images in highly scattering environments, and to permit the investigation of much shorter acquisition time scans. These algorithms take advantage of the data sparseness under the Discrete Cosine Transform (DCT) and the correlation between adjacent pixels, to restore the depth and reflectivity images.

Spectral classification of sparse photon depth images

By illumination of target scenes using a set of different wavelengths, we demonstrate color classification of scenes, as well as depth estimation, in photon-starved images. The spectral signatures are classified with a new advanced statistical image processing method from measurements of the same scene, in this case using combinations of 33, 16, 8 or 4 different wavelengths in the range 500 - 820 nm. This approach makes it possible to perform color classification and depth estimation on images containing as few as one photon per pixel, on average. Compared to single wavelength imaging, this approach improves target discrimination by extracting more spectral information, which, in turn, improves the depth estimation since this approach is robust to changes in target reflectivity. We demonstrate color classification and depth profiling of complex targets at average signal levels as low as 1.0 photons per pixel from as few as 4 different wavelength measurements.

Robust spectral unmixing of multispectral Lidar waveforms

This paper presents a new Bayesian spectral unmixing algorithm to analyse remote scenes sensed via multispectral Lidar measurements. To a first approximation, each Lidar waveform consists of the temporal signature of the observed target, which depends on the wavelength of the laser source considered and which is corrupted by Poisson noise. When the number of spectral bands is large enough, it becomes possible to identify and quantify the main materials in the scene, on top of the estimation of classical Lidar-based range profiles. Thanks to its anomaly detection capability, the proposed hierarchical Bayesian model, coupled with an efficient Markov chain Monte Carlo algorithm, allows robust estimation of depth images together with abundance and outlier maps associated with the observed 3D scene. The proposed methodology is illustrated via experiments conducted with real multispectral Lidar data.

Analysis of three-dimensional scenes using photon-starved data in cluttered target scenarios (Conference Presentation)

This paper investigates a new computational method for reconstruction and analysis of complex 3D scenes. In the presence of targets, Lidar waveforms usually consist of a series of peaks, whose positions and amplitudes depend on the distances of the targets and on their reflectivities, respectively. Inferring the number of surfaces or peaks, as well as their geometric and colorimetric properties becomes extremely difficult when the number of detected photons is low (e.g., short acquisition time) and the ambient illumination is high. In this work, we adopt a Bayesian approach to account for the intrinsic spatial organization of natural scenes and regularise the 3D reconstruction problem. The proposed model is combined with an efficient Markov chain Monte Carlo (MCMC) method to reconstruct the 3D scene, while providing measures of uncertainty (e.g., about target range and reflectivity) which can be used for subsequent decision making processes, such as object detection and recognition. Despite being an MCMC method, the proposed approach presents a competitive computational cost when compared to state-of-the-art optimization-based reconstruction methods, while being more robust to the lack of detected photons (empty or non-observed pixels). Moreover, it includes a multi-scale strategy which allows a quick recovery of coarse approximations of the 3D structures, while is often sufficient for object detection/recognition. We assess the performance of our approach via extensive experiments conducted with real, long-range (hundreds of meters) single-photon Lidar data. The results clearly demonstrate its benefits to infer complex scene content from extremely sparse photon counts.

Time-correlated single-photon counting for single and multiple wavelength underwater depth imaging (Conference Presentation)

A scanning depth imaging system is used for the investigation of three-dimensional image reconstruction and classification of targets in underwater environments. The system uses the Time-Correlated Single-Photon Counting (TCSPC) technique to measure single-photon time-of-flight. In this paper, we use both single and multiple wavelengths to interrogate underwater targets. This presentation will show laboratory measurements on several target scenarios, including targets in clutter. We demonstrate high resolution depth and intensity image reconstruction in highly scattering underwater scenarios, and show image reconstruction at up to nine attenuation lengths between transceiver and target. The system comprised a scanning transceiver unit, fiber coupled to a silicon single-photon avalanche diode (Si SPAD) and a supercontinuum laser system operating at the repetition rate of 19.5 MHz. An acousto-optic tunable filter (AOTF) is used to select an individual operational wavelength in the range 500 nm to 725 nm. The measurements used a range of system configurations, including both single wavelength and multiple wavelength measurements. Generally, the measurements used sub-milliwatt average optical power levels. Bespoke algorithms were developed to identify man-made objects hidden by marine vegetation in the scanned scene. Advanced statistical image processing methods were used to improve target discrimination and to reconstruct the target under different conditions, including reduced number of wavelengths and number of pixels, and reduced acquisition time. Particular attention will be given to the photon starved regime, which will be typical of data acquired at long distances in open ocean waters or in highly scattering environments.