Individual Pixels Research Articles

Deep Learning-based Near-real-time Monitoring of Autumn Irrigation Extent at Sub-pixel Scale in a Large Irrigation District

Flood irrigation is widely applied in harvested croplands of arid regions and can be classified as autumn/winter irrigation (AI) depending on the time of application. Due to its high water consumption, real-time monitoring of the AI extent is crucial to improve its scheduling. We proposed a remote sensing-based long short-term memory (LSTM) model for near-real-time monitoring of AI extent at sub-pixel scale. The model loosely coupled MODIS data with Sentinel-2 data to solve the mixed pixel issue of MODIS data, and calibrated Sentinel-2 thresholds for AI identification by a random forest (RF) module to extract large-scale reference data with high temporal frequencies. The variable importance estimated by RF is used as a reference for feature screening in the LSTM model. As Sentinel-2 images are not available daily, LSTM models trained with incomplete sequences were validated using multiple validation approaches. The model was applied to the Hetao Irrigation District, the largest irrigation district in arid region of China, and delivered reasonable performance with coefficient of determination of over 0.82 and mean absolute error of around 10.7%. Classification of irrigation patterns using simulated time series of irrigation area fractions revealed eight different irrigation patterns from 2010 to 2020 in the study region. Results indicate that the maximum fraction of AI area upstream is closely related to the cropland distribution pattern. The AI patterns changed significantly over the 11 years, with a more pronounced reduction in irrigation duration for individual pixels in the downstream area. These changes are associated with two important land policies implemented in many regions of China, land consolidation and land transfer. The proposed model demonstrates the potential for near-real-time monitoring of autumn irrigation extent within large irrigation districts, which can aid in AI scheduling and provide insight into irrigation patterns and practices.

Local performance evaluation of AI-algorithms with the generalized spatial recall index

Abstract We have developed a novel metric to gauge the performance of artificial intelligence (AI) or machine learning (ML) algorithms, called the Spatial Recall Index (SRI). The novelty is the spatial resolution of a standard performance indicator, as a Recall value is assigned to each individual pixel. This generates a distribution of the performance of a given AI-algorithm with the resolution of the images in the dataset. While the mathematical basis has already been presented before, here we demonstrate the usage on more datasets and delve into in-depth application examples. We examine both the MS COCO and the Berkeley Deep Drive datasets, using a state-of-the-art object detection algorithm. The dataset is degraded using a physical-realistic lens-model, where the optical performance varies over the field of view, as a real camera would. This study highlights the usefulness of the SRI, as every image has been taken by realistic optics. A generalization, the GSRI is introduced, from which we derive SRIA, weighting with object area and SRIrisk intended for autonomous driving. Finally, these metrics are compared.

Potential for remote TD-NIRS imaging using a TCSPC camera

Advances in the CMOS process enables single-photon detection and timing to occur within an individual pixel. Pixel arrays incorporating this in-pixel timing methodology have been increasing in size, facilitating time of flight (ToF) imaging. This means that a fully parallel approach can be made to the collection of images, dramatically speeding up data collection in relation to sample scanning and allowing imaging based on time-correlated single-photon counting (TCSPC) to be made in real time. In this work we make use of a commercial TCSPC camera, based on an array of 192 × 126 imaging pixels with in-pixel timing applied in the area of time-domain near infrared spectroscopy. The potential of remote imaging using this methodology was demonstrated using two model systems. The effect of increasing scattering coefficient using a liquid based (milk added to water) system to assess real time potential was investigated. Additionally a more complex system consisting of a solid with liquid interior (egg) to verify the potential of remote imaging into the interior of the object with different illumination conditions was studied. Using a time-slicing approach intensity images related to each time bin in the ToF distribution were obtained and contrast between the yolk and albumin within the egg was observed.

Pixel-wise Guidance for Utilizing Auxiliary Features in Monte Carlo Denoising

Auxiliary features such as geometric buffers (G-buffers) and path descriptors (P-buffers) have been shown to significantly improve Monte Carlo (MC) denoising. However, recent approaches implicitly learn to exploit auxiliary features for denoising, which could lead to insufficient utilization of each type of auxiliary features. To overcome such an issue, we propose a denoising framework that relies on an explicit pixel-wise guidance for utilizing auxiliary features. First, we train two denoisers, each trained by a different auxiliary feature (i.e., G-buffers or P-buffers). Then we design our ensembling network to obtain per-pixel ensembling weight maps, which represent pixel-wise guidance for which auxiliary feature should be dominant at reconstructing each individual pixel and use them to ensemble the two denoised results of our denosiers. We also propagate our pixel-wise guidance to the denoisers by jointly training the denoisers and the ensembling network, further guiding the denoisers to focus on regions where G-buffers or P-buffers are relatively important for denoising. Our result and show considerable improvement in denoising performance compared to the baseline denoising model using both G-buffers and P-buffers. The source code is available at https://github.com/qbhan/GuidanceMCDenoising.

Improving spatially-resolved lithium quantification in drill core samples of spodumene pegmatite by using laser-induced breakdown spectroscopy and pixel-matched reference areas

Laser ablation-inductively coupled plasma-time of flight mass spectrometry (LA-ICP-TOFMS) concentrations were used to develop accurate calibration models for laser-induced breakdown spectroscopy (LIBS) mappings of pegmatitic drill cores samples. Both methods were applied on the same area of drill core samples, providing two spatially-resolved datasets for this area. The datasets were aligned pixel by pixel to create a pixel-matched reference area that covered the heterogeneity of the complete drill core. This way, different matrix effects affecting LIBS intensities could be taken into account and accurate spatial quantification of Li2O, SiO2, Al2O3, Na2O, and K2O from LIBS measurements was enabled.In particular, LIBS intensities and LA-ICP-TOFMS concentrations of individual pixels of the reference area were used as the input for a linear Partial Least Square Regression (PLSR) and a non-linear Least Square Support Vector Machines (LS-SVM) calibration model. Varying numbers between 100 and 2000 pixels were used for model creation, and root mean square error (RMSE) and R2 of each model were compared. Better values were achieved for the LS-SVM calibration model. Based on these results, the PLSR model was discarded and only the LS-SVM model with 1000 train pixels was further validated.For two different validation areas, LA-ICP-TOFMS concentrations were compared to LIBS-based concentrations obtained from the LS-SVM calibration model. The spatially-resolved quantification results of the LIBS data agree very well with the independently analysed LA-ICP-TOFMS concentrations, which is e.g. reflected in R2 values between 0.83 and 0.96 (mean 0.89) for Li2O, SiO2, Al2O3, Na2O, and K2O. In general, it was shown that the combination of LA-ICP-TOFMS and LIBS yields a comprehensive dataset for robust multivariate calibration. This enables spatially-resolved LIBS-based quantification of pegmatite samples that are subject to significant physical and chemical matrix effects.

Comparative Study Between Vision Transformer and EfficientNet on Marsh Grass Classification

Due to rapidly changing ecosystems, effective environmental protection often calls for the monitoring of the vegetation for any environmental changes. Vegetation monitoring is essential in assessing the changes and impacts to environmentally valuable ecosystems such as marshlands. While vegetation monitoring of marsh grasses is crucial to the maintenance and protection of marshlands, it is a tedious and time-consuming task that involves careful examination of individual pixels within large resolution images. In this study we compare the use of Vision Transformers (ViT) and two different EfficientNet models on automated marsh grass identification using the GTMNERR Marsh Grass Species data set. Our results show that the use of a ViT allowed for an increase in the accuracy of marsh grass identification. The Vision Transformer was also able to better distinguish between the 6 classes in the data set and provided competitive training time to the smaller of the two EfficientNet models tested in this study.

AUTHENTICITY AND THE POOR IMAGE IN THE AGE OF DEEP LEARNING

Deep learning techniques are increasingly used to automate categorization and identification tasks for large datasets of digital photographs. For rasterized images formats, such as JPEGs, GIFs, and PNGs, the analysis happens on the level of individual pixels. Given this, digital images used in deep learning applications are typically restricted to relatively low-resolution formats to conform to the standards of popular pre-trained neural networks. Using Hito Steyerl’s conception of the ‘poor image’ as a theoretical frame, this article investigates the use of these relatively low-resolution images in automated analysis, exploring the ways in which they may be deemed preferable to higher-resolution images for deep learning applications. The poor image is rich in value in this context, as it limits the undesirable ‘noise’ of too much detail. In considering the case of automated art authentication, this article argues that a notion of authenticity is beginning to emerge that raises questions around Walter Benjamin’s often-cited definition in relation to mass image culture. Copies or reproductions are now forming the basis for a new model of authenticity, which exists latently in the formal properties of a digital image.

Large-Scale Terahertz Sensor Array Module With Antenna-Coupled Microbolometers on Glass Substrate With Sigma–Delta ADC Readout ASIC

We describe the design, fabrication and demonstration of a terahertz sensor array module, designed to operate over 250-350GHz. The antenna coupled microbolometer was fabricated on a glass substrate with an integrated array of pixel switches and row multiplexing circuitry implemented with LTPS (Low Temperature PolySilicon) thin film transistors. While other microbolometer arrays have been reported for the upper Terahertz bands, we focus on lower frequencies where adequate semiconductor sources are available potentially enabling important medical and security applications. The devices were fabricated in a Gen 3.5 flat panel manufacturing facility implementing what we call Large Area MEMS Platform (LAMP) surface micromachining process, providing a path to low-cost manufacturing. Individual pixels were defined on a 0.7mm pitch across an array of 128x128 pixels in this first demonstration. Since flat panel fabrication capabilities use large, single-shot reticle masks and large glass substrates, physically larger arrays can be manufactured without mask stitching. An ROIC (readout integrated circuit) was demonstrated to support the THz sensor panel with 648 input channels, each providing an independent sigma delta ADC with excellent uniformity and 12 effective number of bits. An early demonstration of the sensor array is described, imaging a metallic knife hidden in a cardboard envelope.

Topology optimization for isotropic elastic materials using honeycomb tessell

Topology optimization is gaining popularity as a primary tool for engineers in the initial stages of design. Essentially, the design domain is broken down into individual pixels, with the material density of each element or mesh point serving as a design variable. The optimization problem is then tackled through mathematical programming and optimization methods that rely on analytical gradient calculation. In this study, topology optimization using honeycomb tessellation elements is explored. Hexagonal elements have the ability to flexibly connect two adjacent elements. The use of the hexagonal element limits the occurrence of the checkerboard pattern to the finite elements of the quadrilateral standard Lagrangian type. A mathematical model is developed with the objective function being the minimum compliance value of the design domain. The element stiffness matrix is constructed using the strain-displacement matrix and the constitutive matrix, assuming a unit Young's modulus. Additionally, optimal conditions are established using Lagrangian multipliers. Two sensitivity and density filtering filters are employed to increase optimization efficiency, prevent the algorithm from reaching a local optimal state, and speed up convergence. If the suggested filter is employed, the objective function achieves a value of c=173,0293 and convergence is attained after 200 iterations. In contrast, without using the filter, the objective function has a larger value (c=186,7922) and convergence occurs at the 27th iteration. The results are significant for optimizing topology to meet specific boundary condition requirements. This paper proposes a novel approach using a combination of filters to advance topology optimization using hexagonal elements in future applications.

A numerical exploration of hyporheic zone solute transport behavior estimated from electrical resistivity inversions

An increasing number of studies have combined saline tracer injections with time-lapse electrical resistivity (ER) surveys to explore solute exchange within the hyporheic zone. The reliability and resolution of ER surveys of the hyporheic zone are rarely discussed. Here, we use a numerical modeling approach to assess how ER imaging resolves changes in saline tracer concentration within the hyporheic zone given different synthetic scenarios. We create a 3-D synthetic stream and surrounding hyporheic zone with an ER electrode transect and solve the coupled equations for porous fluid flow and solute transport. Then, we solve for the 3-D conduction of current between electrodes to create synthetic time-lapse ER surveys and invert the simulated resistances to obtain 2-D bulk electrical conductivity (σb) images perpendicular to the stream, which we compare with the known synthetic σb fields. The σb fields in the inversion generally capture the shape of the simulated σb fields, but with smoothing and artifacts as a function of the inversion process. Consequently, the calculated cross-sectional area of tracer plume is inaccurate. At inversion pixels beneath the stream, the accuracy of the inverted σb breakthrough curves when compared to the synthetic “truth” varies with stream size, pixel depth, and to a lesser extent injection time. The tails of these pixel breakthrough curves beneath the stream are consistently underestimated compared to the synthetic “truth,” i.e., σb in the inversions appear to return to background faster by one to six hours. The time series of average apparent bulk electrical conductivity requires no inversion and captures lingering saline tracer better than bulk EC breakthrough curves at individual pixels in the inversions.

Developing global annual land surface phenology datasets (1982–2018) from the AVHRR data using multiple phenology retrieval methods

Land surface phenology that reflects periodical changes in terrestrial vegetation plays an important role in influencing the functioning and services of ecosystems. Satellite-derived metrics of land surface phenology have been widely used to assess the patterns in land cover dynamics. There are still limited satellite-based datasets of global land surface phenology available to date due to uneven qualities of satellite data. Here, we developed a global annual land surface phenology dataset based on Advanced Very High Resolution Radiometer (AVHRR) data from 1982 to 2018. The number of global vegetation growing seasons for each individual pixel was counted for each year from 1982 to 2018. We implemented four phenology retrieval methods, including the amplitude threshold method, the second-order derivative method, the third-order derivative method, and the curvature change rate method, to retrieve four key phenological metrics for each vegetation growth cycle, including the start of the growing season (SOS), maturity, senescence, and the end of the growing season (EOS). The results show that the phenological metrics retrieved using different phenology retrieval methods have similar patterns of global spatial distribution. From 1982 to 2018, the timing of both maturity and senescence within a year advanced in the Northern Hemisphere and the timing of SOS delayed in the Southern Hemisphere. The phenological metrics extracted from in-situ data such as the USA National Phenology Network (USA-NPN), the Pan European Phenology Database (PEP725), and flux tower measurements were positively correlated with satellite-derived phenological metrics. Despite the effects of mismatching spatial scales, the retrieved phenological metrics based on 5 km AVHRR data were consistent with those based on 500 m Moderate-resolution Imaging Spectroradiometer (MODIS) data and 30 m fused satellite data at both site-scale and regional-scale. The developed dataset released for usages and applications on global land surface dynamics offers potential end users options for choosing desirable phenological metrics derived from different methods.

Convolutional Neural Network Shows Greater Spatial and Temporal Stability in Multi-Annual Land Cover Mapping Than Pixel-Based Methods

Satellite imagery is the only feasible approach to annual monitoring and reporting on land cover change. Unfortunately, conventional pixel-based classification methods based on spectral response only (e.g., using random forests algorithms) have shown a lack of spatial and temporal stability due, for instance, to variability between individual pixels and changes in vegetation condition, respectively. Machine learning methods that consider spatial patterns in addition to reflectance can address some of these issues. In this study, a convolutional neural network (CNN) model, U-Net, was trained for a 500 km × 500 km region in southeast Australia using annual Landsat geomedian data for the relatively dry and wet years of 2018 and 2020, respectively. The label data for model training was an eight-class classification inferred from a static land-use map, enhanced using forest-extent mapping. Here, we wished to analyse the benefits of CNN-based land cover mapping and reporting over 34 years (1987–2020). We used the trained model to generate annual land cover maps for a 100 km × 100 km tile near the Australian Capital Territory. We developed innovative diagnostic methods to assess spatial and temporal stability, analysed how the CNN method differs from pixel-based mapping and compared it with two reference land cover products available for some years. Our U-Net CNN results showed better spatial and temporal stability with, respectively, overall accuracy of 89% verses 82% for reference pixel-based mapping, and 76% of pixels unchanged over 33 years. This gave a clearer insight into where and when land cover change occurred compared to reference mapping, where only 30% of pixels were conserved. Remaining issues include edge effects associated with the CNN method and a limited ability to distinguish some land cover types (e.g., broadacre crops vs. pasture). We conclude that the CNN model was better for understanding broad-scale land cover change, use in environmental accounting and natural resource management, whereas pixel-based approaches sometimes more accurately represented small-scale changes in land cover.

Uncertainty-guided joint attention and contextual relation network for person re-identification

Due to the influence of factors such as camera angle and pose changes, some salient local features are often suppressed in person re-identification tasks. Moreover, many existing person re-identification methods do not consider the relation between features. To address these issues, this paper proposes two novel approaches: (1) To solve the problem of being confused and misidentified when local features of different individuals have similar attributes, we design a contextual relation network that focuses on establishing the relationship between local features and contextual features, so that all local features of the same person both contain contextual information. (2) To fully and correctly express key local features, we propose an uncertainty-guided joint attention module. The module focuses on the joint representation of individual pixels and local spatial features to enhance the credibility of local features. Finally, our method achieves competitive performance on four widely recognized datasets compared with state-of-the-art methods.

Equalizing the Pixel Response of the Imaging Photoelectric Polarimeter Onboard the IXPE Mission

The Gas Pixel Detector is a gas detector, sensitive to the polarization of X-rays, currently flying onboard the Imaging X-ray Polarimetry Explorer (IXPE)—the first observatory dedicated to X-ray polarimetry. It detects X-rays and their polarization by imaging the ionization tracks generated by photoelectrons absorbed in the sensitive volume, and then reconstructing the initial direction of the photoelectrons. The primary ionization charge is multiplied and ultimately collected on a finely pixellated ASIC specifically developed for X-ray polarimetry. The signal of individual pixels is processed independently and gain variations can be substantial, of the order of 20%. Such variations need to be equalized to correctly reconstruct the track shape, and therefore its polarization direction. The method to do such equalization is presented here and is based on the comparison between the mean charge of a pixel with respect to the other pixels for equivalent events. The method is shown to finely equalize the response of the detectors onboard IXPE, allowing a better track reconstruction and energy resolution, and can in principle be applied to any imaging detector based on tracks.

SIGNAL TRANSMISSION OPTIMIZATION MOVING-VECTOR CLASSIFICATION MPEG STANDARD VIDEO CODING

The rise of useful multimedia applications has sparked a rush to perfect video compression methods. Over the last decade, there has been a lot of interest in using motion-compensated frame interpolation (MCFI) methods to boost video frame rate during playback in both the academic and consumer electronics sectors. Interpolation refers to the process of defining the values of a function at places that are not represented by any of the examples. It accomplishes this process by fitting a constant function across the discrete input samples. We based our methodology on the fact that color cameras function similarly to the human eye. We have been used to black and white photographs, yet the real world is full of color. Grayscale images were the sole option until the advent of color monitors and cameras. Energy assessment calls for photo capture using a camera. Light's electromagnetic waves provide the observable energy in this scheme. For motion-compensated video vector frame interpolation or frame rate up-conversion and also Phase frame interpolation Method, the authors present a novel approach with low vector dispensation method in the conclusion. Using optical flow or other conventional methods, it is necessary to have precise pixel correspondences between pictures in order to compute interpolated frames in a video series. We provide a practical replacement by capitalizing on advances in phase-based approaches, which use the phase shift of individual pixels to express motion. In challenging interpolation conditions, such as major appearance changes, flow-based algorithms frequently create severe visual distortions, yet our solution fails gently. Our approach is particularly useful for retiming and interpolating frames in high-resolution video shot at a high frame rate. Estimating motion for the purpose of improving the quality of the decompressed video requires more research on the nature of the compression problem. a novel way to estimating motion that is analogous to full-search 3ss,4ss, block-matching, and parametric techniques. The primary focus will be research into both error-free compression and the development of novel motion estimation algorithms. The results of the research indicate that the suggested system has the potential to increase visual superiority and is also robust, especially when applied to video sequences with quick movements and challenging portions. By combining phase-based and conventional Lagrangian, which is utilized to get better quality results and deal with variations in light more gracefully.

Signal transmission optimization moving-vector classification Mpeg standard video coding

The rise of useful multimedia applications has sparked a rush to perfect video compression methods. Over the last decade, there has been a lot of interest in using motion-compensated frame interpolation (MCFI) methods to boost video frame rate during playback in both the academic and consumer electronics sectors. Interpolation refers to the process of defining the values of a function at places that are not represented by any of the examples. It accomplishes this process by fitting a constant function across the discrete input samples. We based our methodology on the fact that color cameras function similarly to the human eye. We have been used to black and white photographs, yet the real world is full of color. Grayscale images were the sole option until the advent of color monitors and cameras. Energy assessment calls for photo capture using a camera. Light's electromagnetic waves provide the observable energy in this scheme. For motion-compensated video vector frame interpolation or frame rate up-conversion and Phase frame interpolation Method, the authors present a novel approach with low vector dispensation method in the conclusion. Using optical flow or other conventional methods, it is necessary to have precise pixel correspondences between pictures in order to compute interpolated frames in a video series. We provide a practical replacement by capitalizing on advances in phase-based approaches, which use the phase shift of individual pixels to express motion. In challenging interpolation conditions, such as major appearance changes, flow-based algorithms frequently create severe visual distortions, yet our solution fails gently. Our approach is particularly useful for retiming and interpolating frames in high-resolution video shot at a high frame rate. Estimating motion for the purpose of improving the quality of the decompressed video requires more research on the nature of the compression problem. a novel way to estimating motion that is analogous to full-search 3ss,4ss, block-matching, and parametric techniques. The primary focus will be research into both error-free compression and the development of novel motion estimation algorithms. The results of the research indicate that the suggested system has the potential to increase visual superiority and is also robust, especially when applied to video sequences with quick movements and challenging portions. By combining phase based and conventional Lagrangian, which is utilized to get better quality results and deal with variations in light more gracefully.

Signal transmission optimization moving-vector classification mpeg standard video coding

The rise of useful multimedia applications has sparked a rush to perfect video compression methods. Over the last decade, there has been a lot of interest in using motion-compensated frame interpolation (MCFI) methods to boost video frame rate during playback in both the academic and consumer electronics sectors. Interpolation refers to the process of defining the values of a function at places that are not represented by any of the examples. It accomplishes this process by fitting a constant function across the discrete input samples. We based our methodology on the fact that color cameras function similarly to the human eye. We have been used to black and white photographs, yet the real world is full of color. Grayscale images were the sole option until the advent of color monitors and cameras. Energy assessment calls for photo capture using a camera. Light's electromagnetic waves provide the observable energy in this scheme. For motion-compensated video vector frame interpolation or frame rate up-conversion and Phase frame interpolation Method, the authors present a novel approach with low vector dispensation method in the conclusion. Using optical flow or other conventional methods, it is necessary to have precise pixel correspondences between pictures to compute interpolated frames in a video series. We provide a practical replacement by capitalizing on advances in phase-based approaches, which use the phase shift of individual pixels to express motion. In challenging interpolation conditions, such as major appearance changes, flow-based algorithms frequently create severe visual distortions, yet our solution fails gently. Our approach is particularly useful for retiming and interpolating frames in high-resolution video shot at a high frame rate. Estimating motion for the purpose of improving the quality of the decompressed video requires more research on the nature of the compression problem. a novel way to estimating motion that is analogous to full-search 3ss,4ss, block-matching, and parametric techniques. The primary focus will be research into both error-free compression and the development of novel motion estimation algorithms. The results of the research indicate that the suggested system has the potential to increase visual superiority and is also robust, especially when applied to video sequences with quick movements and challenging portions. By combining phase based and conventional Lagrangian, which is utilized to get better quality results and deal with variations in light more gracefully.

Speckle Pattern Improvements for Digital Image Correlation to Capture Small Strains in Tensile Tests on Rock

Digital image correlation (DIC) is a popular optical strain measurement technology for deformation analysis in rock mechanics laboratory testing. Although reliable and repeatable, traditional strain gauge and extensometer measurement techniques have drawbacks including high recurring cost and limited spatial distribution of data. DIC utilizes continuous photography and computer software to track paint speckle pixel displacements of a material deforming under load, producing a full-field strain map for each recorded increment. Although DIC presents an opportunity to overcome the limitations of traditional methods, significant precision and care is required to track typical total displacements of 0.005 mm (0.01% strain) in Brazilian tensile stress (BTS) tests of igneous and sedimentary specimens measuring 47.6 mm diameter. For computer algorithms to precisely track individual pixels between frames, it is often necessary to apply a unique speckle pattern conforming to several criteria to the rock specimen. A variety of speckle application techniques have been reported in rock mechanics literature; however, the importance of a high-quality speckle pattern and resulting influence on the accuracy of DIC results is frequently overlooked. The laboratory testing program in this research utilized 2-Dimensional DIC on BTS tests to investigate the cost, efficiency, ease of application, and effectiveness for several speckle application methods. In addition to testing traditional spraypaint, airbrush, and stamp application methods, an innovative laser engravement speckle application technique was developed in this research that provides more reliable DIC results. This research provides practical speckle pattern application recommendations for small strain 2D DIC measurements.

Imaging effects due to pixel distortions in CdZnTe detectors—results from the HREXI Calibration Facility

ProtoEXIST2 (P2) was a prototype imaging X-ray detector plane developed for wide-field Time Domain Astrophysics (TDA) in the 5 - 200 keV energy band. It was composed of an 8 $\times$ 8 array of 5 mm thick, 2cm $\times$ 2cm pixelated (32 $\times$ 32) CdZnTe (CZT) detectors with a 0.6 mm pitch that utilize the NuSTAR ASIC(NuASIC) for readout. During the initial detector development process leading up to post-flight examination of the entire detector plane, distortions in expected pixel positions and shapes were observed in a significant fraction of the detectors. The HREXI (High Resolution Energetic X-ray Imager) Calibration Facility (HCF) was designed and commissioned to improve upon these early experiments and to rapidly map out and characterize pixel non-uniformities and defects within CZT detector planes at resolutions down to 50 $\rm \mu$m. Using this facility, the sub-pixel level detector response of P2 was measured at 100 $\rm \mu$m resolution and analyzed to extract and evaluate the area and profile of individual pixels, their morphology across the entire P2 detector plane for comparison with previous measurements and to provide additional characterization. In this article, we evaluate the imaging performance of a coded-aperture telescope using the observed pixel morphology for P2 detectors. This investigation will serve as an initial guide for detector selection in the development of HREXI detector planes, for the future implementation of the 4pi X-Ray Imaging Observatory (4piXIO) mission which aims to provide simultaneous and continuous imaging of the full sky ($\rm 4\pi$ sr) in the 3-200 keV energy band with $\rm \simeq$ 2 arcmin angular resolution and $\simeq$ 10 arcsec source localization, as well as other, future coded-aperture instruments.

Scatter correction for cone-beam CT via scatter kernel superposition-inspired convolutional neural network

Objective. X-ray scatter leads to signal bias and degrades the image quality in Computed Tomography imaging. Conventional real-time scatter estimation and correction methods include the scatter kernel superposition (SKS) methods, which approximate x-ray scatter field as a convolution of the scatter sources and scatter propagation kernels to reflect the spatial spreading of scatter x-ray photons. SKS methods are fast to implement but generally suffer from low accuracy due to the difficulties in determining the scatter kernels. Approach. To address such a problem, this work describes a new scatter estimation and correction method by combining the concept of SKS methods and convolutional neural network. Unlike conventional SKS methods which estimate the scatter amplitude and the scatter kernel based on the value of an individual pixel, the proposed method generates the scatter amplitude maps and the scatter width maps from projection images through a neural network, from which the final estimated scatter field is calculated based on a convolution process. Main Results. By incorporating physics in the network design, the proposed method requires fewer trainable parameters compared with another deep learning-based method (Deep Scatter Estimation). Both numerical simulations and physical experiments demonstrate that the proposed SKS-inspired convolutional neural network outperforms the conventional SKS method and other deep learning-based methods in both qualitative and quantitative aspects. Significance. The proposed method can effectively correct the scatter-related artifacts with a SKS-inspired convolutional neural network design.