Search

Cilj urbanističkog rješenja bio je osigurati novo atraktivno poslovno središte koje će prizvati velik broj korisnika. Navedeno postižemo ostvarenjem osnovnog koncepta koji se temelji na principima pikselizacije, gradacije i ozelenjavanju prostora. Točkastu izgradnju različitih visina, raspoređujemo po parceli u ozelenjen prostor, te tako stvaramo kvalitetne prostore u zelenilu, ali i atraktivno središte koje je smješteno na ulazu (jednom od ulaza) u grad Osijek, čime također doprinosimo i slici garada. Ideja arhitektonskog rješenja proizlazi upravo iz urbanističkog koncepta. Osnovni je element pri oblikovanju poslovne zgrade piksel uz nadodanu treču dimenziju=kocka, koja se u skladu s okolnim okruženjem spiralno uzdiže uz konstantnu vertikalu (jezgru) u mreži konstrukcije. Stvarajući na taj način zanimljiv i dinamičan odnos puno-prazno principima adicije i redukcije, uz dodatak vertikalnog zelenila, koje se stupovima uspinje po cijeloj visini zgrade., The goal of the urbanism solution was to ensure a new attractive business center that will appeal to a large number of users. We achieved the stated by realizing the basic concept which is established on the principles of pixelization, gradation and greening of space. We arrange the point construction of different heights on the plot in a green space, thus create quality spaces in the greenery, but also an attractive center located at the entrance (one of the entrances) to the city of Osijek, which also contributes to the image of the city. The idea of an architectural solution derives precisely from the urban concept. The basic element in the design of an office building is a pixel with an added third dimension = a cube, which in accordance with the surrounding environment spirally rises with a constant vertical (core) in the network of the structure. Thus creating an interesting and dynamic full-empty relationship to the principles of addition and reduction, with the addition of vertical greenery, which rises in columns along the entire height of the building.

The best crystal identification (CI) algorithms proposed so far for phoswich detectors are based on adaptive filtering and pulse shape discrimination (PSD). However, these techniques require free running analog to digital converters, which is no longer possible with the ever increasing pixelization of new detectors. We propose to explore the dual-threshold time-over-threshold (ToT) technique, used to measure events energy and time of occurence, as a more robust solution for crystal identification with broad energy windows in phoswich detectors. In this study, phoswich assemblies made of various combinations of LGSO and LYSO scintillators with decay times in the range 30 to 65 ns were investigated for the LabPET II detection front-end. The electronic readout is based on a 4 × 8 APD array where pixels are individually coupled to charge sensitive preamplifiers followed by first order CR-RC shapers with 75 ns peaking time. Crystal identification data were sorted out based on the measurements of likeliness between acquired signals and a time domain model of the analog front-end. Results demonstrate that crystal identification can be successfully performed using a dual-threshold ToT scheme with a discrimination accuracy of 99.1% for LGSO (30 ns)/LGSO (45 ns), 98.1% for LGSO (65 ns)/LYSO (40 ns) and 92.1% for LYSO (32 ns)/LYSO (47 ns), for an energy window of [350–650] keV. Moreover, the method shows a discrimination accuracy >97% for the two first pairs and ~90% for the last one when using a wide energy window of [250–650] keV.

Rad predstavlja projekt poslovne zgrade u idejnom urbanističkom rješenju Poslovne zone Gacka, na križanju Trpimirove i Gacke ulice na jugu Osijeka. Projektirana zgrada je lamela čija se duža pročelja okreću se planiranoj stambenoj namjeni i javnom trgu unutar Zone. Dvije komunikacijske jezgre pružaju jednak odnos i uvlačenje korisnika s obje strane, a unutrašnji prostor gradira se od najvećeg i zajedničkog na jugu do privatnih ureda na sjeveru. Jezgre ujedno služe kao zona diferencijacije različitih karaktera radnih prostora., This paper presents the project of a business building in the proposed urban solution of Gacka business zone, located on the intersection of Trpimirova and Gacka Street, in the south part of Osijek. The designed building is longitudinally shaped with its fronts facing the planned residential area and the public square. Two communication cores offer the equal treatment to users from both sides. Internal space is formed by gradation from mutual open-space offices in the south to private offices in the northern end of the building. The cores create differentiation zones that separate working spaces of contrast characteristics.

Zadatak završnog rada je urbanističko i arhitektonsko rješenje poslovne zone Gacka koja se smjestila u industrijsku zonu na južnom dijelu grada Osijeka. Obuhvat je omeđen Ulicom Kneza Trpimira te novoplaniranom ulicom sa zapadne strane. U urbanističkom projektu, na parceli je planirana poslovna i ugostiteljsko-turistička namjena - tri poslovne zgrade različitih tipologija te hotel na sjeveroistočnom dijelu obuhvata. Na parceli su određene privatne poslovne površine, ali i javne površine koje se vraćaju gradu na korištenje. Za tri poslovne zgrade izvedene su dvije podzemne garaže. Urbanističko rješenje nastalo je iz analize postojećeg stanja - tipologiji okolne izgradnje, postojećih vizura, analizi povijesnog razvoja grada Osijeka te očekivanih kretanja ljudi. Arhitektonsko rješenje izvedeno je za kvadratnu zgradu na jugozapadnom dijelu obuhvata. Zgrada je projektirana za rad na ekološko-poljoprivrednoj proizvodnji. Njen koncept odgovor je na početne ideje utemeljene u urbanističkom rješenju te u analizi parcele - zgrada kao ''hala'', pikselizacija te analiza očekivanih kretanja i pristupa ljudi. Zgrada sadržava privatne prostore za rad, administraciju i istraživanje zaposlenika, multifunkcionalni prostor za izložbe i ostale razne namjene te javne društvene prostore. Prostori su organizirani u dva volumena zgrade koji su povezani mostovima, a uz unutarnje prostore sadržava i razne vanjske prostore različitih namjena. Na ravan krov ove kvadratne zgrade postavljeni su solarni paneli. Zgrada je karakteristična po kontrastu svog velikog volumena nasuprot lakoći cijele svoje pojavnosti te ''ulici'' koja se ostvaruje po dijagonali tlocrta., The task of the final paper is the urban and architectural solution of the business zone Gacka, which is located in the industrial zone in the southern part of the city of Osijek. The coverage is bordered by Kneza Trpimira Street and the newly planned street on the west side. In the urban project, business and catering-tourist purpose is planned on the plot - three business buildings of different typologies and a hotel in the north-eastern part of the scope. There are certain private business areas on the plot, but also public areas that are returned to the city for use. Two underground garages have been constructed for three business buildings. The urban solution arose from the analysis of the existing situation - the typology of the surrounding construction, the existing views, the analysis of the historical development of the city of Osijek and the expected movements of people. The architectural solution was made for a square building in the south-western part of the scope. The building is designed for work on organic-agricultural production. Her concept is a response to the initial ideas based in the urban solution and in the analysis of the plot - the building as a "hall", pixalization and analysis of expected movements and approaches to people. The building contains private spaces for work, administration and research of employees, a multifunctional space for exhibitions and other various purposes, and public social spaces. The spaces are organized into two volumes of the building which are connected by bridges, and in addition to the interior spaces it also contains various outdoor spaces for different purposes. Solar panels are placed on the flat roof of this square building. The building is characterized by the contrast of its large volume as opposed to the lightness of its entire appearance and the "street" that is realized on the diagonal of the floor plan.

Urbanističko rješenje proizlazi iz tematike pikselizacije položenog u zelenilo, iščitavajući okolnu urbanu matricu. Projekt zgrade se sastoji od jasnog vertikalnog zoninga, odnosno podjele na javno, polujavno i privatno. Uz sam naglasak vertikane podjele u doprinos prilaže i promjena konstrukcije. Polujavni dio zamišljen kao mjesto za odmor na radnom mjestu, ali i za posjetitelje izložbenih prostora., The urban solution derives from the theme of pixelization laid out in the greenery, reading the surrounding urban matrix. The project of the building consists of a clear vertical zoning, ie division into public, semi-public and private. In addition to the emphasis on the vertical division, the change in construction also contributes. Semi-public part conceived as a place to relax in the workplace, but also for visitors to exhibition spaces.

This letter introduces a systematic procedure to reduce the mutual coupling (MC) between radiating elements in a microstrip array antenna. The proposed method is based on using parasitic elements between the radiating elements. The shape of the parasitic element, which impacts the MC and other radiation properties of the antenna, is determined in a systematic pattern optimization procedure based on pixelization of the area between the antennas and application of a binary optimization algorithm. A bit with binary values of 1 or 0 is assigned to each pixel, which signifies the presence or absence of copper on the pixel surface, respectively. Afterwards, during a binary particle swarm optimization algorithm, the optimal value of each bit is obtained, which results in the optimum shape of the parasitic element. In the optimization process, reducing the MC between the radiating elements is set as the main optimization objective while maintaining good antenna impedance matching and preserving of the radiation pattern. Using the proposed method, we designed a parasitic decoupling element (resonator) between two microstrip patch antennas, which were placed close to each other. The results demonstrate that adding this resonator between the two antennas fulfilled the design objectives and reduced the MC between radiating elements by 24 dB, while preserving the radiation pattern and impedance matching. The results of the simulations have been verified by fabrication and measurements.

Nowadays, a company uses many sensors to record its entire activity process; the recorded data are called event-log. However, event-log prevalently contains discrete data that many powerful machine-learning algorithms are unable to deal with. One-hot encoding is an outstanding method for transforming discrete data into a binary vector. Nonetheless, if there are many distinct values, the problem of dimensionality will be incurred. To tackle this issue, we propose a new approach, called the Pixelization method, which transforms event data into images. We experimentally performed causal inference for prediction of pixels (representing the processing time of each event) by using a generative model with our novel convolution technique. We compared our approach with a baseline method, one-hot encoding, and an entity-embedded approach combined with a neural network model. The results showed that our approach outperforms the state-of-the-art methods in terms of accuracy.

In this work, we present a sound mathematical result that may correct the pixelization bias in the calculation of the line edge roughness (LER) even when the pixel size of SEM images is much larger than its rms value. This result can be useful in the LER metrology of current and future lithographic patterns where the desired rms value of LER is in the deep sub-nm range and smaller than the used pixel sizes. We computationally justify the predictions of the mathematical result and then demonstrate its application in synthesized SEM images where we can separate the impact of noise and focus on the pixelization effects in LER measurements. Furthermore, this result emphasizes the power of mathematical and computational methods to enhance the metrological output of SEM images in deep nanometer scale.

The Python Sky Model (PySM) is a Python package used by Cosmic Microwave Background (CMB) experiments to simulate maps, in HEALPix pixelization, of the various diffuse astrophysical components of Galactic emission relevant at CMB frequencies (i.e. dust, synchrotron, free-free and Anomalous Microwave Emission), as well as the CMB itself. These maps may be integrated over a given instrument bandpass and smoothed with a given instrument beam. PySM 2, released in 2016, has become the de-facto standard for simulating Galactic emission, for example it is used by CMB-S4, Simons Observatory, LiteBird, PICO, CLASS, POLARBEAR and other CMB experiments, as shown by the 80+ citations of the PySM 2 publication. As the resolution of upcoming experiments increases, the PySM 2 software has started to show some limitations, the solution to these issues was to reimplement PySM from scratch focusing on these features: reimplement all the models with the numba Just-In-Time compiler for Python to reduce memory overhead and optimize performance; use MPI through mpi4py to coordinate execution of PySM 3 across multiple nodes and rely on libsharp, for distributed spherical harmonic transforms; employ the data utilities infrastructure provided by astropy to download the input templates and cache them when requested. At this stage we strive to maintain full compatibility with PySM 2, therefore we implement the exact same astrophysical emission models with the same naming scheme. In the extensive test suite we compare the output of each PySM 3 model with the results obtained by PySM 2., Submitted to JOSS, Github repository: https://github.com/galsci/pysm

The complex interplay of magnetohydrodynamics, gravity, and supersonic turbulence in the interstellar medium (ISM) introduces non-Gaussian structure that can complicate comparison between theory and observation. We show that the Wavelet Scattering Transform (WST), in combination with linear discriminant analysis (LDA), is sensitive to non-Gaussian structure in 2D ISM dust maps. WST-LDA classifies magnetohydrodynamic (MHD) turbulence simulations with up to a 97\% true positive rate in our testbed of 8 simulations with varying sonic and Alfv\'{e}nic Mach numbers. We present a side-by-side comparison with two other methods for non-Gaussian characterization, the Reduced Wavelet Scattering Transform (RWST) and the 3-Point Correlation Function (3PCF). We also demonstrate the 3D-WST-LDA and apply it to classification of density fields in position-position-velocity (PPV) space, where density correlations can be studied using velocity coherence as a proxy. WST-LDA is robust to common observational artifacts, such as striping and missing data, while also sensitive enough to extract the net magnetic field direction for sub-Alfv\'{e}nic turbulent density fields. We include a brief analysis of the effect of point spread functions and image pixelization on 2D-WST-LDA applied to density fields, which informs the future goal of applying WST-LDA to 2D or 3D all-sky dust maps to extract hydrodynamic parameters of interest., Comment: 21 pages, 13 figures, submitted to ApJ

Depth perception is an important building block for user experience in AR/VR/MR applications. It is based on visual cues such as blurring, sharpness, relative size, non-uniform details level, and more. The human brain uses these visual cues to create a 3D perception. Displaying enhanced 3D cues requires high resolution while maintaining a large Field of View (FOV). However, current display technologies fall short of fulfilling these requirements. Current solutions require sophisticated optics and loads of additional hardware. Previous work by the authors presented a novel method to produce a foveated image using a Laser-Based Scanning (LBS) display enabling a localized high resolution and large FOV solution for enhanced 3D cues. This work takes this further and presents an implementation that enables a multifocal image projection combined with a foveated display. Multifocal plains provide natural blurring and sharpness cues. The solution is comprised of multi degrees of freedom scanning MEMS mirrors, a laser projector, and an algorithm enabling foveated projection and changeable focal plains. As the focal plains could be changed from frame to frame, the combination of the created image along with foveated pixelization generates a more natural display. This allows visualizing enhanced depth cues which significantly improves the 3D perception and 3D user experience. Moreover, this setup enabled reduced form factor. The authors believe that the presented solution will enable a state-of-the-art user experience with an acceptable form factor that will enable commercialization into daily use AR/VR devices.

In this paper, we focus on the across-edges pixelization effects on LER measurement accuracy. The pixelization across edges rounds the detected edge position according to pixel size and is expected to impact the accuracy of LER measurements (rms, correlation length, PSD curve). Given the similar size of targeted LER values in IRDS and of the pixel size used in SEM measurements, these effects are expected to gain an increasing interest and therefore to be worth studying. Our investigation is utilized with the aid of synthesized images (characterized by complete control of roughness and image parameters), with the final aim being the exploration of the link between the selected pixel size of the SEM image and the observed deviation of the measured rms (rms_m) from the true rms value (rms_t). More specifically, the goal is to find the role of the ratio of pixel size to rms_t in the measurement of rms and how we can predict it.

An analysis of the source position differences between VLBI-based ICRF and $Gaia$-CRF catalogues is a key step in assessing their systematic errors and determining their mutual orientation. One of the main factors that limits the accuracy of determination of the orientation parameters between two frames is the impact of outliers. To mitigate this effect, a new method is proposed based on pixelization data over the equal-area cells, followed by median filtering of the data in each cell. After this, a new data set is formed, consisting of data points near-uniformly distributed over the sphere. The vector spherical harmonics (VSH) decomposition is then applied to this data to finally compute the orientation parameters between ICRF and $Gaia$ frames. To validate the proposed approach, a comparison was made of the ICRF3-SX and $Gaia$~DR2 catalogues using several methods for outliers removal. The results of this work showed that the proposed method is practically insensitive to outliers and thus provides much more robust results of catalogues comparison than the methods used so far. This conclusion was confirmed by analogous test comparison of the $Gaia$~DR2 and OCARS catalogues., Comment: Accepted in MNRAS

A new class of silicon detectors, the Low-Gain Avalanche Diode (LGAD), has already shown excellent timing performances O(30-40 ps) and is being considered for several applications in the most diverse fields, ranging from timing in high energy physics experiments to medical imaging. LGADs typically exhibit a low to moderate gain (5 - 100), short rise time and high signal-to-noise ratio. Fine pixelization of LGADs is difficult to achieve, and to provide fine spatial resolution the AC-coupled LGAD (AC-LGAD) approach was introduced. In this type of device, the signal is capacitively induced on fine-pitched electrodes placed over an insulator. LGAD and AC-LGAD prototypes have been designed and fabricated at Brookhaven National Laboratory and segmented in both pixel matrices and strips. These prototypes are characterized via Transient Current Technique (TCT) using a fast-pulsed IR laser. The timing resolution is measured by studying the time coincidence between signals generated by a beta particle beam from a 90Sr source in two sensors or by using a collimated proton beam. Additional studies of the sensor's performance have been carried out, including the induced signal in neighbouring pixels and strips.

For some patients with advanced retinal degenerative diseases, retinal prostheses are used to partially restore vision currently. However, due to limited number of implantable electrodes, the visual perception produced by the implants is in low resolution and has limited information bandwidth. Computer vision is the key to overcome the limitations and to optimize the information assignment. As a result, many researchers devote to investigating image processing algorithms to optimize the visual percepts of recipients. In this study, we focus on complex scene perception and present a new approach to build a hierarchical presentation. An object detection network (Mask R-CNN) is used for foreground objects extraction and segmentation. Then, a post-processing method referring to human visual attention mechanism is applied to generate this hierarchical presentation. Psychophysical experiments prove that under simulated prosthetic vision, the proposed hierarchical image processing strategy has prominent advantages in complex scene perception comparing with direct pixelization, especially when overlapping occurs in the scene.

Protecting the identity of users in social networks and photographic media has become one of the most important concerns, especially with the breakthrough of internet and information technology. Thus several techniques, such as blurring and pixelization, have been introduced by privacy-protection technology designers to make the features of a user’s face unrecognizable. However, many researchers pointed out the ineffectiveness of these techniques. In this paper with deal with the problem of reconstructing obfuscated human faces. We propose a Deep Neural Network based on U-Net and trained on Labeled Faces in the Wild (LFW) dataset. Our model is evaluated using two image quality metrics PSNR and SSIM. Experimental results show the ability of our model to reconstruct images that have underwent different forms of obfuscation (blurring and pixelization) with great accuracy, outperforming previous work.

This paper presents a monocular camera-based lead vehicle distance and velocity estimation algorithm for automotive application. With an initial guess of real width of the lead vehicle, a Kalman Filter gives estimates for relative distance, velocity and acceleration. The still unknown scale factor to the real size is then statistically estimated from multiple hypothesis using vertical triangulation measurements. Camera pitching motion effects are compensated through the estimation of the vanishing point. The real relative distance, velocity and acceleration can be obtained with the estimated scale factor. The developed method is evaluated in simulations considering the Euro NCAP forward collision warning and emergency braking test procedures, the braking dynamics of vehicles, multiple lead vehicle sizes, periodic camera pitching disturbance, pixelization and vanishing point estimation errors and a wide range of velocities from 10km/h to 130km/h. The results are promising and so real life evaluation is the goal of future development.

In this paper, a systematic procedure to reduce mutual coupling between linear $2 \times 1$ array microstrip antennas is proposed. The proposed design procedure is based on pixelization of the surface between two patch elements and the application of a binary particle swarm optimization (BPSO) algorithm. In the binary optimization algorithm, each pixel is represented by a bit that gets a binary value of 1 or 0 indicating the presence or absence of copper on the area of the pixel. The optimization goal is set to minimizing the mutual coupling between two elements while the impedance matching and radiation pattern are not degraded. The concept of the work is validated by full-wave simulations and measurements.

Non-planar photovoltaics (NP-PV) are an emerging application space enabled by advances in thin-film photovoltaic (PV) materials. However, there are challenges to the practical use due to the complexities introduced with a non-planar PV surface. Curved surfaces have built-in partial shading and even self-shading - phenomenon that leads to non-uniform current densities. The conditioning electronics responsible for the maximizing production is highly dependent on how the cells are connected. Specifically, maximum power point tracking performance is contingent on the impact partial shading has on the circuit. Shaded cells in a series connected systems inhibit the flow of current which reduces the output power and can cause irreparable damage to the material. Since non-planar self-shading can be predicted, an optimal configuration that can reduce current variance while increasing usable output power is possible. By analyzing the production patterns of the surface, this paper presents a methodology, called pixelization, to characterize and design an optimal connection scheme with for non-planar applications. Pixelization is the grouping of similarly producing cells for the purpose of increasing energy conversion efficacy. The results of this work give insight into how geometry and surface symmetry affect how non-planar can be designed to maximize the potential of the geometry.

Spatial cross-matching operation over geospatial polygonal datasets is a highly compute-intensive yet an essential task to a wide array of real-world applications. At the same time, modern computing systems are typically equipped with multiple processing units capable of task parallelization and optimization at various levels. This mandates for the exploration of novel strategies in the geospatial domain focusing on efficient utilization of computing resources, such as CPUs and GPUs. In this paper, we present a CPU-GPU hybrid platform to accelerate the cross-matching operation of geospatial datasets. We propose a pipeline of geospatial subtasks that are dynamically scheduled to be executed on either CPU or GPU. To accommodate geospatial datasets processing on GPU using pixelization approach, we convert the floating point-valued vertices into integer-valued vertices with an adaptive scaling factor as a function of the area of minimum bounding box. We present a comparative analysis of GPU enabled cross-matching algorithm implementation in CUDA and OpenACC accelerated C++. We test our implementations over Natural Earth Data and our results indicate that although CUDA based implementations provide better performance, OpenACC accelerated implementations are more portable and extendable while still providing considerable performance gain as compared to CPU. We also investigate the effects of input data size on the IO / computation ratio and note that a larger dataset compensates for IO overheads associated with GPU computations. Finally, we demonstrate that an efficient cross-matching comparison can be achieved with a cost-effective GPU.

In this paper, we present a novel unsupervised learning method for pixelization. Due to the difficulty in creating pixel art, preparing the paired training data for supervised learning is impractical. Instead, we propose an unsupervised learning framework to circumvent such difficulty. We leverage the dual nature of the pixelization and depixelization, and model these two tasks in the same network in a bi-directional manner with the input itself as training supervision. These two tasks are modeled as a cascaded network which consists of three stages for different purposes. GridNet transfers the input image into multi-scale grid-structured images with different aliasing effects. PixelNet associated with GridNet to synthesize pixel arts with sharp edges and perceptually optimal local structures. DepixelNet connects the previous network and aims to recover the pixelized result to the original image. For the sake of unsupervised learning, the mirror loss is proposed to hold the reversibility of feature representations in the process. In addition, adversarial, L1, and gradient losses are involved in the network to obtain pixel arts by retaining color correctness and smoothness. We show that our technique can synthesize crisper and perceptually more appropriate pixel arts than state-of-the-art image downscaling methods. We evaluate the proposed method with extensive experiments on many images. The proposed method outperforms state-of-the-art methods in terms of visual quality and user preference.

Background and objective Current retinal prostheses can only generate low-resolution visual percepts constituted of limited phosphenes which are elicited by an electrode array and with uncontrollable color and restricted grayscale. Under this visual perception, prosthetic recipients can just complete some simple visual tasks, but more complex tasks like face identification/object recognition are extremely difficult. Therefore, it is necessary to investigate and apply image processing strategies for optimizing the visual perception of the recipients. This study focuses on recognition of the object of interest employing simulated prosthetic vision. Method We used a saliency segmentation method based on a biologically plausible graph-based visual saliency model and a grabCut-based self-adaptive-iterative optimization framework to automatically extract foreground objects. Based on this, two image processing strategies, Addition of Separate Pixelization and Background Pixel Shrink, were further utilized to enhance the extracted foreground objects. Results i) The results showed by verification of psychophysical experiments that under simulated prosthetic vision, both strategies had marked advantages over Direct Pixelization in terms of recognition accuracy and efficiency. ii) We also found that recognition performance under two strategies was tied to the segmentation results and was affected positively by the paired-interrelated objects in the scene. Conclusion The use of the saliency segmentation method and image processing strategies can automatically extract and enhance foreground objects, and significantly improve object recognition performance towards recipients implanted a high-density implant.

Machine learning, as a computational tool for finding mappings between the input and output data, has been widely used in engineering fields. Researchers have applied machine learning models to generate 2D drawings with pixels or 3D models with voxels, but the pixelization reduces the precision of the geometries. Therefore, in order to learn and generate 3D geometries as vectorized models with higher precision and faster computation speed, we develop a specific artificial neural network, learning and generating design features for the forms of buildings. A customized data structure with feature parameters is constructed, meeting the requirements of the neural network by rebuilding surfaces with controlling points and appending additional input neurons as quantified vectors to describe the properties of the design. The neural network is first trained with generated design data and then tested by adjusting the feature parameters. The prediction of the generated data shows the basic generative ability of the neural network. Furthermore, trained with design data collected from existing buildings, the neural network learns and infers the geometric design features of architectural design with different feature parameters, providing a data-driven method for designers to generate and analyze architectural forms.

Deep Neural Networks (DNNs) are known to be susceptible to adversarial examples. Adversarial examples are maliciously crafted inputs that are designed to fool a model, but appear normal to human beings. Recent work has shown that pixel discretization can be used to make classifiers for MNIST highly robust to adversarial examples. However, pixel discretization fails to provide significant protection on more complex datasets. In this paper, we take the first step towards reconciling these contrary findings. Focusing on the observation that discrete pixelization in MNIST makes the background completely black and foreground completely white, we hypothesize that the important property for increasing robustness is the elimination of image background using attention masks before classifying an object. To examine this hypothesis, we create foreground attention masks for two different datasets, GTSRB and MS-COCO. Our initial results suggest that using attention mask leads to improved robustness. On the adversarially trained classifiers, we see an adversarial robustness increase of over 20% on MS-COCO.

Cadmium–zinc–telluride (CZT) pixel detectors represent a consolidated choice for the development of room-temperature spectroscopic X-ray imagers, finding important applications in medical imaging, often as detection modules of a variety of new SPECT and CT systems. Detectors with 3–5 mm thicknesses are able to efficiently detect X-rays up to 140 keV giving reasonable room-temperature energy resolution. In this work, the room-temperature performance of 3 mm-thick CZT pixel detectors, recently developed at IMEM/CNR of Parma (Italy), is presented. Sub-millimetre detector arrays with pixel pitch less than 500 µm were fabricated. The detectors are characterized by good room-temperature performance even at high bias voltage operation (6000 V cm−1), with energy resolutions (FWHM) of 3% (1.8 keV) and 1.6% (2 keV) at 59.5 keV and 122.1 keV, respectively. Charge-sharing investigations were performed with both uncollimated and collimated synchrotron X-ray beams with particular attention to recovering the charge losses at the inter-pixel gap region. High rate measurements demonstrated the absence of high-flux radiation-induced polarization phenomena up to 25 × 106 photons mm−2 s−1.

In this paper, a systematic procedure to reduce mutual coupling between linear 2×1 array microstrip antennas is proposed. The proposed design procedure is based on pixelization of the surface between two patch elements and application of a binary particle swarm optimization (BPSO) algorithm. In the binary optimization algorithm, each pixel is represented by a bit which gets a binary value of 1 or 0 indicating the presence or absence of copper on the area of the pixel. The optimization goal is set to minimizing the mutual coupling between two elements while the impedance matching and radiation pattern are not degraded. Full-wave simulation results are provided.

Providing an overview of an ensemble of oil reservoir models could help users compare and analyze their characteristics. Approaches that show a single model at a time may hamper analysts’ understanding of the whole model set. In this paper, we propose two visualization approaches that show multiple reservoir models, simultaneously and on a single screen, with the goal of helping users to compare models and improve their understanding of ensemble characteristics. First, we calculate 2D models from the ensemble's 3D models. We then create two visualizations that represent ensembles of these 2D models. The Small Multiples approach lays out heatmaps of 2D models side-by-side on a grid. Pixelization approach shows n 2D models in a single heatmap, where each cell (i, j) contains n subcells that represent values in the coordinate (i, j) of each model. Both approaches display their elements (heatmaps and subcells) clustered by X-means, which may help analysts identify similarities and representative models in the ensemble. We used two types of distance matrices: based on Euclidean distance of models for a given property or, based on Euclidean distance of feature vectors of the 2D models. We tested our approaches within models based on Brazilian benchmark cases corresponding to a turbiditic reservoir (UNISIM-I-D/M/H) and a presalt carbonatic reservoir (UNISIM-II-D). As a result, the Small Multiples approach presented clusters of similar models for some properties of the ensembles we studied, e.g. eight clusters of porosity values in UNISIM-II-D's ensemble. This fact suggests that eight representative models can represent the ensemble, regarding porosity. Also, a Pixelization approach revealed patterns that happen in specific regions of all models of an ensemble, such as an abrupt change of porosity values in the northwest region of UNISIM-I-M's models. Both approaches have the potential to help analysts perceive situations that would be improbable to observe in a graph with only mean values for each cell. Therefore, our proposal can be helpful to users who need to deal with uncertainties and have an overview of ensembles of models for better understanding and decisionmaking, e.g. when they need to choose representative models for a process of decision analysis related to petroleum field development and management.

A new method SREAG (spherical rectangular equal-area grid) is proposed to divide a spherical surface into equal-area cells. The method is based on dividing a sphere into latitudinal rings of near-constant width with further splitting each ring into equal-area cells. It is simple in construction and use, and provides more uniform width of the latitudinal rings than other methods of equal-area pixelization of a spherical surface. The new method provides a rectangular grid cells with the latitude- and longitude-oriented boundaries, near-square cells in the equatorial rings, and the closest to uniform width of the latitudinal rings as compared with other equal-area isolatitudinal grids. The binned data is easy to visualize and interpret in terms of the longitude-latitude rectangular coordinate system, natural for astronomy and geodesy. Grids with arbitrary number of rings and, consequently, wide and theoretically unlimited range of cell size can be built by the proposed method. Comparison with other methods used in astronomical research showed the advantages of the new approach in sense of uniformity of the ring width, a wider range of grid resolution, and simplicity of use., Accepted in AJ. Supporting Fortran routines are available at http://www.gaoran.ru/english/as/ac_vlbi/index.htm#SREAG

Aiming at the deficiencies of the current face privacy protection based on image editing in terms of protection capability and image utility, a novel face privacy protection scheme using CNN based ROI (Region of interest) editing is proposed. In this paper, a novel face ROI calculation method based on CNN (convolutional neural Network) is first put forward. A set of CNN networks is selected and utilized to calculate the gradient of each face image in an image dataset, and then an average gradient image is obtained from the image gradients. After that, the sub-ROIs are determined from the average gradient image, and different sub-ROIs have different contribution to the output of the CNN. Based on face ROI editing, a privacy protection scheme is put forward, which is adaptive to the existing privacy protection methods such as blurring, pixelization, encryption and etc. Experimental results and analysis show that it can achieve the same privacy protection capability with less modification and better image quality compared with the existing methods.

Improvements as well as new functionalities can be implemented in IR sensors by depositing a stack of thin film layers on top their surface. For this purpose, Safran Reosc has developed and qualified various processes, from conventional deposition methods to planar structuration approaches. In this article, the usefulness of thin film AR coatings are addressed and the latest results obtained on projects like Sentinel-5 and Microcarb IR focal planes are presented and discussed. We also address our progress in pixelization of the coating by structuration of these optical layers. Eventually, the most recent developments of sub-lambda photonic filtering structures are discussed.

In this paper we presented the new algorithm for pixelization of astronomical data in celestial sphere. This algorithm is used for partition and analysis of the large astronomical catalogues that contain data of up to billion objects. The proposed algorithm for pixelization can be applied for indexing of the objects positions in the celestial sphere. This allows the quick search in the large astronomical databases. The search results of indexed objects can be used by different consumers such as robotic telescopes, ground and flying robots that need to guidance, navigate and even to track the situation in celestial sphere. The problems of sphere partition to pixels by different methods are also described in the paper.

We construct an overcomplete and multiscale dictionary of bandlimited Slepian functions on the sphere. Slepian functions are the bandlimited eigenfunctions obtained by solving spatial-spectral concentration problem on the sphere. To this end, we develop the hierarchical equal area iso-latitude iso-longitude pixelization (HEALLPix) scheme for hierarchical partitioning of the sphere into equal area sub-regions called pixels and present its quaternary tree structure. We then solve the concentration problem of finding bandlimited functions with maximal energy concentration in the given spatial region for each pixel and use these spatially concentrated bandlimited functions as dictionary elements. We analyze the span of the dictionary elements and their mutual coherence and show that the dictionary spans the space of bandlimited functions which are optimally (energy) concentrated within a pixel on the sphere with most of its elements exhibiting negligibly small mutual coherence. Hence, the proposed dictionary is a significant tool for use in multi-resolution analysis and sparse reconstruction of signals on the sphere.

Fluctuations in the large-scale structure of the Universe contain significant information about cosmological physics, but are modulated in survey datasets by various observational effects. Building on existing literature, we provide a general treatment of how fluctuation power spectra are modified by a position-dependent selection function, noise, weighting, smoothing, pixelization and discretization. Our work has relevance for the spatial power spectrum analysis of galaxy surveys with spectroscopic or accurate photometric redshifts, and radio intensity-mapping surveys of the sky brightness temperature including generic noise, telescope beams and pixelization. We consider the auto-power spectrum of a field, the cross-power spectrum between two fields and the multipoles of these power spectra with respect to a curved sky, deriving the corresponding power spectrum models, estimators, errors and optimal weights. We note that "FKP weights" for individual tracers do not in general provide the optimal weights when measuring the cross-power spectrum. We validate our models using mock datasets drawn from N-body simulations, and provide the python code we use for these tests at https://github.com/cblakeastro/intensitypower. Our treatment should be useful for modelling and studying cosmological fluctuation fields in observed and simulated datasets., 15 pages, 2 figures, accepted for publication by MNRAS

Spatial cross-matching operation over geospatial polygonal datasets is important to a variety of GIS applications. However, it involves extensive computation cost associated with intersection and union of a geospatial polygon pair from large scale datasets. This mandates for exploration of parallel computing capabilities such as GPU to increase the efficiency of such operations. In this paper, we present a CPU-GPU hybrid platform to accelerate the cross-matching operation of geospatial datasets. The computing tasks are dynamically scheduled to be executed either on CPU or GPU. To accommodate geospatial datasets processing on GPU using pixelization approach, we convert the floating point-valued vertices into integer-valued vertices with an adaptive scaling factor as a function of area of minimum bounding box. We test our framework over Natural Earth Dataset and achieve 10x speedup on NVIDIA GeForce GTX750 GPU and 14x speedup on Tesla K80 GPU over 280,000 polygon pairs in one tile and 400 tiles in total. We also investigate the effects of input data size to the IO / computation ratio and note that the sufficiently large input data size is required to better utilize the computing power of GPU. Finally, with comparison between two GPUs, our results demonstrate that the efficient cross-matching comparison can be achieved with a cost-effective GPU.

Our goal is to explore novel techniques for blurring real-world objects with a physical filter that changes their appearance and naturally induces a line of sight. To reach this goal, we have developed a system, called "Fade-in Pixel", that enables real-world objects to be pixelized gradually and selectively. We realized this gradual and selective pixelization by using 3D printing a transparent lens array filter and controlling the filling and discharging of transparent liquid having the same refractive index as that of the transparent resin. The result is that objects in the real world appear pixelized when the liquid is discharged and appear clear when the liquid is filled. This paper describes the motivation for this study, along with the design, implementation, and interactive applications of Fade-in Pixel.

In this work, we report a direct (non-iterative) algorithm to reconstruct the three-dimensional (3D) momentum-space picture of any charged particles collected with a velocity-map imaging system from the two-dimensional (2D) projected image captured by a position-sensitive detector. The method consists of fitting the measured image with the 2D projection of a model 3D velocity distribution defined by the physics of the light-matter interaction. The meaningful angle-correlated information is first extracted from the raw data by expanding the image with a complete set of Legendre polynomials. Both the particle's angular and energy distributions are then directly retrieved from the expansion coefficients. The algorithm is simple, easy to implement, fast, and explicitly takes into account the pixelization effect in the measurement.

This study estimates the coefficient of friction on the actual curling sheet with a pebble, without having to mount any devices on it. Existing studies have equipped the ice sheet with external devices to estimate the coefficient of friction. Another paper estimated friction coefficients, assuming that the pebble did not exist, because it was very difficult to estimate the coefficient of friction in the space where the pebble was located. In this study, it is possible to estimate the coefficient of friction of the actual curling sheet without fitting an external device onto the curling sheet. In addition, the friction coefficient estimation method attempted using image processing techniques may be most similar to the actual friction coefficient, reducing the outside influence, compared to the existing papers. It will also be an estimation method using image processing techniques, which form the basis of various friction coefficient screening methods.

We propose an iso-latitude sampling scheme for the representation of band-limited signals on the sphere. The proposed scheme is designed as a variant of the widely used Hierarchical Equal Area iso-Latitude Pixelization (HEALPix) scheme on the sphere. We use HEALPix grid of resolution $N_{\mathrm{side}}$ to represent a signal of bandlimit (spherical harmonic degree) $L$ . To reduce the number of samples, the proposed algorithm takes only $L$ iso-latitude rings out of $4N_{\mathrm{side}}-1$ rings of the HEALPix. This selection is carried out by ensuring that the spherical harmonic transform (SHT) of the signal is computed accurately from the samples. The number of samples required by the proposed sampling scheme is smaller than that required by HEALPix by at least a factor of 3/2. For the proposed sampling scheme, we also formulate the spherical harmonic transform and conduct numerical experiments to evaluate the number of samples required by the proposed sampling scheme and the accuracy of the associated SHT.