Learning a Probabilistic Strategy for Computational Imaging Sensor Selection
California Institute of Technology（加州理工学院）；Harvard Medical School（哈佛医学院）；Massachusetts Institute of Technology（麻省理工学院）
He Sun, Adrian V. Dalca, Katherine L. Bouman
备注：
This paper has been accepted to the IEEE International Conference on Computational Photography (ICCP) 2020. Keywords: Computational Imaging, Optimized Sensing, Ising Model, Deep Learning, VLBI, Interferometry
Optimized sensing is important for computational imaging in low-resource environments, when images must be recovered from severely limited measurements. In this paper, we propose a physics-constrained, fully differentiable, autoencoder that learns a probabilistic sensor-sampling strategy for optimized sensor design. The proposed method learns a system's preferred sampling distribution that characterizes the correlations between different sensor selections as a binary, fully-connected Ising model. The learned probabilistic model is achieved by using a Gibbs sampling inspired network architecture, and is trained end-to-end with a reconstruction network for efficient co-design. The proposed framework is applicable to sensor selection problems in a variety of computational imaging applications. In this paper, we demonstrate the approach in the context of a very-long-baseline-interferometry (VLBI) array design task, where sensor correlations and atmospheric noise present unique challenges. We demonstrate results broadly consistent with expectation, and draw attention to particular structures preferred in the telescope array geometry that can be leveraged to plan future observations and design array expansions.
Computational image reconstruction algorithms generally produce a single image without any measure of uncertainty or confidence. Regularized Maximum Likelihood (RML) and feed-forward deep learning approaches for inverse problems typically focus on recovering a point estimate. This is a serious limitation when working with underdetermined imaging systems, where it is conceivable that multiple image modes would be consistent with the measured data. Characterizing the space of probable images that explain the observational data is therefore crucial. In this paper, we propose a variational deep pr…
High dynamic range (HDR) imaging is one of the biggest achievements in modern photography. Traditional solutions to HDR imaging are designed for and applied to CMOS image sensors (CIS). However, the mainstream one-micron CIS cameras today generally have a high read noise and low frame-rate. These, in turn, limit the acquisition speed and quality, making the cameras slow in the HDR mode. In this paper, we propose a new computational photography technique for HDR imaging. Recognizing the limitations of CIS, we use the Quanta Image Sensor (QIS) to trade the spatial-temporal resolution with bit-de…
University of Bonn（波恩大学）；Stanford University（斯坦福大学）
Computational photography encompasses a diversity of imaging techniques, but one of the core operations performed by many of them is to compute image differences. An intuitive approach to computing such differences is to capture several images sequentially and then process them jointly. Usually, this approach leads to artifacts when recording dynamic scenes. In this paper, we introduce a snapshot difference imaging approach that is directly implemented in the sensor hardware of emerging time-of-flight cameras. With a variety of examples, we demonstrate that the proposed snapshot difference ima…
Université catholique de Louvain（天主教鲁汶大学）；Université libre de Bruxelles（布鲁塞尔自由大学）；imec（imec研究中心）
This paper introduces two acquisition device architectures for multispectral compressive imaging. Unlike most existing methods, the proposed computational imaging techniques do not include any dispersive element, as they use a dedicated sensor which integrates narrowband Fabry-Pérot spectral filters at the pixel level. The first scheme leverages joint inpainting and super-resolution to fill in those voxels that are missing due to the device's limited pixel count. The second scheme, in link with compressed sensing, introduces spatial random convolutions, but is more complex and may be affected …
Stanford University（斯坦福大学）；The University of Manchester（曼彻斯特大学）
The ability to image high-dynamic-range (HDR) scenes is crucial in many computer vision applications. The dynamic range of conventional sensors, however, is fundamentally limited by their well capacity, resulting in saturation of bright scene parts. To overcome this limitation, emerging sensors offer in-pixel processing capabilities to encode the incident irradiance. Among the most promising encoding schemes is modulo wrapping, which results in a computational photography problem where the HDR scene is computed by an irradiance unwrapping algorithm from the wrapped low-dynamic-range (LDR) sens…
Adverse weather conditions are very challenging for autonomous driving because most of the state-of-the-art sensors stop working reliably under these conditions. In order to develop robust sensors and algorithms, tests with current sensors in defined weather conditions are crucial for determining the impact of bad weather for each sensor. This work describes a testing and evaluation methodology that helps to benchmark novel sensor technologies and compare them to state-of-the-art sensors. As an example, gated imaging is compared to standard imaging under foggy conditions. It is shown that gate…
Displaced automotive sensor imaging exploits joint processing of the data acquired from multiple radar units, each of which may have limited individual resources, to enhance the localization accuracy. Prior works either consider perfect synchronization among the sensors, employ single antenna radars, entail high processing cost, or lack performance analyses. Contrary to these works, we develop a displaced multiple-input multiple-output (MIMO) frequency-modulated continuous-wave (FMCW) radar signal model under coarse synchronization with only frame-level alignment. We derive Bayesian performanc…
Imaging in low light is difficult because the number of photons arriving at the sensor is low. Imaging dynamic scenes in low-light environments is even more difficult because as the scene moves, pixels in adjacent frames need to be aligned before they can be denoised. Conventional CMOS image sensors (CIS) are at a particular disadvantage in dynamic low-light settings because the exposure cannot be too short lest the read noise overwhelms the signal. We propose a solution using Quanta Image Sensors (QIS) and present a new image reconstruction algorithm. QIS are single-photon image sensors with …
Heriot-Watt University（赫瑞-瓦特大学）；The University of Edinburgh（爱丁堡大学）；Defence Science and Technology Laboratory（国防科学技术实验室）
Single-Photon Avalanche Detector (SPAD) arrays are a rapidly emerging technology. These multi-pixel sensors have single-photon sensitivities and pico-second temporal resolutions thus they can rapidly generate depth images with millimeter precision. Such sensors are a key enabling technology for future autonomous systems as they provide guidance and situational awareness. However, to fully exploit the capabilities of SPAD array sensors, it is crucial to establish the quality of depth images they are able to generate in a wide range of scenarios. Given a particular optical system and a finite im…
Because image sensor chips have a finite bandwidth with which to read out pixels, recording video typically requires a trade-off between frame rate and pixel count. Compressed sensing techniques can circumvent this trade-off by assuming that the image is compressible. Here, we propose using multiplexing optics to spatially compress the scene, enabling information about the whole scene to be sampled from a row of sensor pixels, which can be read off quickly via a rolling shutter CMOS sensor. Conveniently, such multiplexing can be achieved with a simple lensless, diffuser-based imaging system. U…
We demonstrate optics-free imaging of complex QR-codes using a bare image sensor and a trained artificial neural network (ANN). The ANN is trained to interpret the raw sensor data for human visualization. The image sensor is placed at a specified gap from the QR code. We studied the robustness of our approach by experimentally testing the output of the ANNs with system perturbations of this gap, and the translational and rotational alignments of the QR code to the image sensor. Our demonstration opens us the possibility of using completely optics-free cameras for application-specific imaging o…
Superconducting QUantum-Interference Devices (SQUIDs) make magnetic resonance imaging (MRI) possible in ultra-low microtesla-range magnetic fields. In this work, we investigate the design parameters affecting the signal and noise performance of SQUID-based sensors and multichannel magnetometers for MRI of the brain. Besides sensor intrinsics, various noise sources along with the size, geometry and number of superconducting detector coils are important factors affecting the image quality. We derive figures of merit based on optimal combination of multichannel data, analyze different sensor arra…
X-Ray imaging is quick, cheap and useful for front-line care assessment and intra-operative real-time imaging (e.g., C-Arm Fluoroscopy). However, it suffers from projective information loss and lacks vital volumetric information on which many essential diagnostic biomarkers are based on. In this paper we explore probabilistic methods to reconstruct 3D volumetric images from 2D imaging modalities and measure the models' performance and confidence. We show our models' performance on large connected structures and we test for limitations regarding fine structures and image domain sensitivity. We …
Synthetic apertures find applications in many fields, such as radar, radio telescopes, microscopy, sonar, ultrasound, LiDAR, and optical imaging. They approximate the signal of a single hypothetical wide aperture sensor with either an array of static small aperture sensors or a single moving small aperture sensor. Common sense in synthetic aperture sampling is that a dense sampling pattern within a wide aperture is required to reconstruct a clear signal. In this article we show that there exists practical limits to both, synthetic aperture size and number of samples for the application of occl…
Single-photon avalanche diodes (SPADs) are an emerging technology with a unique capability of capturing individual photons with high timing precision. SPADs are being used in several active imaging systems (e.g., fluorescence lifetime microscopy and LiDAR), albeit mostly limited to low photon flux settings. We propose passive free-running SPAD (PF-SPAD) imaging, an imaging modality that uses SPADs for capturing 2D intensity images with unprecedented dynamic range under ambient lighting, without any active light source. Our key observation is that the precise inter-photon timing measured by a S…
Imaging across both the full transverse spatial and temporal dimensions of a scene with high precision in all three coordinates is key to applications ranging from LIDAR to fluorescence lifetime imaging. However, compromises that sacrifice, for example, spatial resolution at the expense of temporal resolution are often required, in particular when the full 3-dimensional data cube is required in short acquisition times. We introduce a sensor fusion approach that combines data having low-spatial resolution but high temporal precision gathered with a single-photon-avalanche-diode (SPAD) array wit…
State-of-the-art time-of-flight (ToF) based 3D sensors suffer from poor lateral and depth resolutions. In this work, we introduce a novel sensor concept that provides ToF-based 3D measurements of real world objects with depth precisions up to 35 micrometers and point cloud densities at the native sensor-resolutions of state-of-the-art CMOS/CCD cameras (up to several megapixels). Unlike other continuous-wave amplitude-modulated ToF principles, our approach exploits wavelength diversity for an interferometric surface measurement of macroscopic objects with rough or specular surfaces. Based on th…
A multispectral image camera captures image data within specific wavelength ranges in narrow wavelength bands across the electromagnetic spectrum. Images from a multispectral camera can extract additional information that the human eye or a normal camera fails to capture and thus may have important applications in precision agriculture, forestry, medicine and object identification. Conventional multispectral cameras are made up of multiple image sensors each fitted with a narrow passband wavelength filter and optics, which makes them heavy, bulky, power hungry and very expensive. The multiple …
Close to half of the world population have smartphones, while a typical flagship smartphone today has been integrated with more than 20 smart components and sensors, making a smartphone a highly integrated platform that can potentially mimic the five senses of humans. Recent advancement in achieving high compactness, high performance computing, high flexibility, and multiplexed functionality in smartphones have enabled them for many cutting-edge healthcare applications, such as single-molecule imaging, medical diagnosis, and biosensing, which were conventionally done with bulky and sophisticat…
Conventional methods of wavefront reconstruction from the raw data of the Shack-Hartmann sensor use the focal spot shifts and discard the high-frequency information about the wavefront. Phase-retrieval-based methods treat the Hartmann pattern as the diffraction image and use the Rayleigh-Sommerfeld propagation to estimate the wavefront with greater accuracy and resolution. In this Letter, we propose a novel approach to the phase-retrieval-based reconstruction by considering the Hartmann pattern as a point-spread function of a general imaging system with an introduced phase diversity of a speci…
We report a new lensless microscopy configuration by integrating the concepts of transverse translational ptychography and defocus multi-height phase retrieval. In this approach, we place a tilted image sensor under the specimen for linearly-increasing phase modulation along one lateral direction. Similar to the operation of ptychography, we laterally translate the specimen and acquire the diffraction images for reconstruction. Since the axial distance between the specimen and the sensor varies at different lateral positions, laterally translating the specimen effectively introduces defocus mu…
Ghost imaging can capture 2D images with a point detector instead of an array sensor. It therefore offers a solution to the challenge of building area format sensors in wavebands where such sensors are difficult and expensive to produce and opens up new imaging modalities due to high-performance single-pixel detectors. Traditionally, ghost imaging retrieves the image of an object offline, by correlating measured light intensities and applied illuminating patterns. Here we present a feedback-based approach for online updating of the imaging result that can bypass post-processing, termed self-ev…
We present the first 1Mpixel SPAD camera ever reported. The camera features 3.8ns time gating and 24kfps frame rate; it was fabricated in 180nm CIS technology. Two pixels have been designed with a pitch of 9.4$μ$m in 7T and 5.75T configurations, respectively, achieving a maximum fill factor of 13.4%. The maximum PDP is 27%, median DCR 2.0cps, variation in gating length 120ps, position skew 410ps, and rise/fall time <550ps, all FWHM at 3.3V of excess bias. The sensor was used to capture 2D/3D scenes over 2m with an LSB of 5.4mm and a precision better than 7.8mm. Extended dynamic range is demons…
To be able to resolve molecular-clusters it is crucial to access vital informations (such as, molecule density and cluster-size) that are key to understand disease progression and the underlying mechanism. Traditional single-molecule localization microscopy (SMLM) techniques use molecules of variable sizes (as determined by its localization precisions (LPs)) to reconstruct super-resolution map. This results in an image with overlapping and superimposing PSFs (due to a wide size-spectrum of single molecules) that degrade image resolution. Ideally it should be possible to identify the brightest …
