Angular Space Research Articles

Normalized Maximal Margin Loss for Open-Set Image Classification

This work aims to address the image classification problem under open-set protocol: classes in test set do not appear in the training set. Intuitively, convolutional Neural Network (CNN) with softmax loss is a straight-forward solution. However, the unknown class (is not predefined in the training set) makes the boundaries of intra-class and inter-class more blurred, which brings more challenges for image classification. Although some softmax variants, such as center loss, CosFace loss etc., focus on learning discriminative features by minimizing the intra-class distance, they do not explicitly maximize inter-class distance, which is more important for open-set problem verified by our experiments. Besides, even though deep metric learning, such as with the contrastive loss and the triplet loss, can learn discriminative features of intra-class and inter-class, it needs a time-consuming image sampling process during training. In this paper, we propose a novel normalized maximal margin (NMM) loss for open-set image classification, which not only explicitly minimizes intra-class distance and maximizes inter-class distance, but also defines their margins. Specially, after analyzing the advantage of angular space that the softmax loss normalized by the feature and weights through geometric interpretation, we make NMM work in angular space. Then, the validity of NMM for discriminative features learning is demonstrated from the view of geometric interpretation as well. After that, we innovatively determine the upper bound of inter-class margin by theoretical analysis. Finally, extensive experiments are conducted on popular datasets: CIFAR-100 (object recognition), ImageNet (image classification), LFW (face recognition) and MSMT17 (person re-identification) to verify the effectiveness of NMM. The experimental results show that NMM achieves very competitive performance.

Quantum interface for high-dimensional quantum states encoded in an orbital angular momentum space

Quantum interface for high-dimensional quantum states encoded in an orbital angular momentum space

A Method for Tailoring the Gain Pattern of a Single Antenna Element

In this paper, we present a novel method to tailor the radiation pattern of an antenna element, such as a single element of an array, by maximizing its partial radiation efficiency in the desired angular space. The tailored pattern of the single element is achieved by current distribution synthesis using multiple feed points in the antenna. A method to determine and realize the currents at the feeds is presented. The method is demonstrated by designing and manufacturing a four-element linear array consisting of tailored antenna elements and comparing its performance to the reference case. The element consists of four closely-spaced metallic patches placed on a substrate backed with a ground plane, and each patch is separately fed. The initial four-port feed is reduced to a single feed with a feeding network based on the determined currents. A conventional patch antenna is chosen as the reference. The simulated and measured results show improvement in the realized gain in the desired angular space.

The Utilization of the Radon Transform for the Extraction of the Orientation of Linear Features in Binary Images

This paper considers the idea of using the Radon transform to extract the orientation features of lines in binary images. The Radon transform sends a line at a particular orientation in image space to a point in feature space. The ensuing set of points in feature space is called a sinogram. This computation is usually performed for a large group of angles (over the interval [0,179] degrees taken in integer increments in this paper). Therefore, linear features at specific orientations will be mapped to points having maximum value at particular angles. For angular spacing of at least 5 degrees, the peaks of the sinogram at the angles corresponding to the orientations of the lines will be clearly visible (in a bar plot of sinogram peaks) above sinogram values at other angles. The mapping from image space to feature space accomplished by the Radon transform which maps rectangular coordinates (x,y) to coordinates (range, angle) provides for the garnering of the orientation of the linear features in binary images. In particular, the coordinates (range, angle) in the sinogram allow for distinguishing between lines oriented at one angle versus lines oriented at another angle or angles. This particular property of the sinogram allows for the extraction of the orientation features of lines in an image.

Effect of geometrical features on the contra-rotating propeller hydrodynamic performances

New contra-rotating four-bladed DTMB propeller configurations operating in open water are numerically studied to determine their hydrodynamic performances. The unsteady turbulent flow around propellers is modelled by RANS equations with Kω SST model then solved by a CFD code. The computational domain is divided in two blocks linked with a rotating interface. The obtained results show that thrust and efficiency of the contra-rotating (CRP) increase compared to the single propeller, leading to a significant reduction of the propeller diameter. The variation in axial spacing and angular displacement seems to have little effect on the CRP efficiency. The results also show that the thrust can be further improved by adopting a moderate negative twist angle of the rear propeller.
 Keywords: CRP contra-rotating propeller, axial spacing, angular spacing, twist angle, CFD.

Using angular momentum maps to detect kinematically distinct galactic components

ABSTRACT In this work, we introduce a physically motivated method of performing disc/spheroid decomposition of simulated galaxies, which we apply to the eagle sample. We make use of the healpix package to create Mollweide projections of the angular momentum map of each galaxy’s stellar particles. A number of features arise on the angular momentum space which allows us to decompose galaxies and classify them into different morphological types. We assign stellar particles with angular separation of less/greater than 30° from the densest grid cell on the angular momentum sphere to the disc/spheroid components, respectively. We analyse the spatial distribution for a subsample of galaxies and show that the surface density profiles of the disc and spheroid closely follow an exponential and a Sérsic profile, respectively. In addition discs rotate faster, have smaller velocity dispersions, are younger and are more metal rich than spheroids. Thus, our morphological classification reproduces the observed properties of such systems. Finally, we demonstrate that our method is able to identify a significant population of galaxies with counter-rotating discs and provide a more realistic classification of such systems compared to previous methods.

Full-sky bispectrum in redshift space for 21cm intensity maps

We compute the tree-level bispectrum of 21cm intensity mapping after reionisation. We work in directly observable angular and redshift space, focusing on equal-redshift correlations and thin redshift bins, for which the lensing contribution is negligible. We demonstrate the importance of the contributions from redshift-space distortions which typically dominate the result. Taking into account the effects of telescope beams and foreground cleaning, we estimate the signal to noise, and show that the bispectrum is detectable by both SKA in single-dish mode and HIRAX in interferometer mode, especially at the lower redshifts in their respective ranges.

Investigation of Low Temperature SiP Epitaxy on 300 mm Si Substrate

To continue the scaling down of transistors several challenges needs to be addressed, including the contact resistivity at the interface of the contact metal and the source/drain (SD) region, and the source/drain bulk resistivity. Phosphorous doped Si (Si:P) is commonly used for nFET as an alternative for (Si:C):P. In this work we report on low temperature (LT) epitaxy (below 500℃) of Si:P layers with extremely low as-grown layer resistivity. Optimal growth conditions at 400℃ led to a layer resistivity of 0.2 mOhm.cm and an active dopant concentration of ~1.15×1021 cm-3. Finally, the as-grown layer properties are compared to existing literature data.SiP layers were grown in a 300 mm ASM Intrepid® ES reduced pressure chemical vapor deposition reactor on Si(001) substrates. The Intrepid® system is equipped with a Previum® module for pre-epi cleaning at low temperatures. First, a set of Si:P layers was epitaxially grown at 470°C, with various Si precursor flows. The pressure, carrier flow and PH3 flow were kept constant. Figure 1 shows the omega-2theta scans around the symmetric (004) reflections of the Si:P/Si stacks. As expected, reducing the Si precursor flow leads to a higher substitutional P incorporation in the layer. The Si:P layer is pseudomorphic with good crystalline quality, as indicated by the thickness fringes. Figure 2 summarizes the HR-XRD angular spacing versus the SIMS content from different literature papers [1-4]. We note that for a low P doping level (<4%) the data are almost overlapping, while the spread increases for a high P content. This could be attributed different SIMS calibration for high P content. Using these calibration curves we note that the P content in the layers shown in Fig. 1 varies between ~1% to around 8.5% ± 0.5%. Figure 3 represents the resistivity versus the P content at three different temperatures (470°C, 440°C and 400°C). Definitely, the growth conditions were tuned to enable a growth with an acceptable growth rate at 400°C. The results indicate that reducing the growth temperature leads to a reduction in the layer resistivity, for the same P content. This highlights that the well-known P deactivation mechanism is correlated to the growth temperature. Figure 4 reveals AFM scans for the Si:P layers with different P content showing that smooth surface morphology can be obtained. Figure 5 compares the (micro-) Hall mobility versus the active concentration for the Si:P process developed in this frame work at LT to data obtained for DCS based process at high temperature (HT) (at 670°C). The DCS based layers were subjected to different thermal anneals [1]. The LT Si:P process follows the same trend, i.e. the Hall mobility decreases when the active dopant concentration increases. The reduction could be attributed to an increase in ionized impurity scattering. This explains the resistivity value around 0.2 mOhm.cm regardless of the P content at 400°C. Figure 6 shows the active dopant concentration as a function of the total dopant concentration. The LT process significantly outperforms the HT process in terms of ratio of active dopant over total dopant concentration. Furthermore, it shows an improvement in active dopant concentration even when compared to the layers exposed to ms laser anneal. Figure 7 summarizes the resistivity value for a Si:P layer obtained at high temperature and subsequently subjected to different anneals and the Si:P layer grown at 400°C, with a P content around 3%. Figure 8 shows the smooth surface morphology of the LT SiP layer with a resistivity of 0.2 mOhm.cm. Finally, Table 1 summarizes the key parameters for the Si:P layers found in literature and compare it to this work.To conclude, LT Si:P epitaxy paves the road towards new applications and is required for novel device architecture. However, a selective process is at utmost importance for integration schemes, which is currently under investigation.[1] E. Rosseel et al, ECS Trans. 75(8) (2016) 347.[2] K.D Weeks et al, Thin Solid Films 520 (2012) 3158.[3] M. Lee et al, ACS Appl. Electron. Mater. 1(3) (2019) 288.[4] X. Li et al, ECS Trans. 64(6) (2014) 959.[5] J.M. Hartmann et al, Semicond. Sci. Technol. 32 (2017) 104003.[6] C.N. Ni et al, VLSI Technology, Kyoto, (2015) T118-T119.[7] M. Bauer el al, ECS Trans. 33(6) (2010) 629.[8] Z. Ye et al, ECS Trans. 50(9) (2012) 1007. Figure 1

Weak feature extraction for early wear of connecting rod bearings under transient conditions

As we all know, it is difficult to extract weak feature for early wear of connecting rod bearings under transient conditions. In order to solve the problem, a method of extracting wear features for connecting rod bearings based on variational modal decomposition (VMD) adaptive noise reduction and computational order tracking (COT) was proposed. Firstly, the vibration signals of internal combustion engine block under transient operating conditions were collected, and the signals were reordered to satisfy the order tracking condition. Then, interpolation and fitting techniques were used to map the signal from the time domain space to the angular domain space. Next, the VMD was used to decompose the angular domain signal into multiple modal components, and the autocorrelation function (ACF) was used to denoise the modal components adaptively. Finally, the signal was reconstructed to conduct COT analysis, and the wear features of connecting rod bearing were extracted by average COT spectrum. The simulation analysis and the simulation experiment of the connecting rod bearing fault show that the proposed method is effective, and the weak features for early wear of connecting rod bearing of internal combustion engine are extracted.

A white-box impersonation attack on the FaceID system in the real world

The arcface model maximizes the classification boundary in the angular space, and is one type of the best models in the current face recognition model. We propose a mask sticker attack method to realize the impersonation attack of arcface model. The method specifically uses a parabolic transformation to simulate the bending situation of a sticker on a mask, and uses a multi-stage PGD attack to generate a adversarial sticker. Finally, the adversarial sticker is attached to the mask worn by the attacker to perform an impersonation attack. On targets of different skin colors, ages and genders, the method proposed in this article has an attack success rate of 0.65, and the final cosine similarity between attacker with mask sticker and target can reach about 0.5-0.7. In addition, we discuss the effect of sticker location and size on attack effect, as well as the generalization of this method on other models.

Two-step system for image receiving in OAM-SK-FSO link.

Turbulence resistance is a significant research area for orbital angular momentum shift keying-based free space optical communication (OAM-SK-FSO). We put forward a two-step combinational system to receive high fidelity image data from the atmospheric turbulence (AT) channels. Firstly, the AT-detector-based multi-CNN (ATDM-CNN) demodulator is proposed which is very different from the traditional single-CNN (S-CNN) demodulator. The AT detector detects the AT strength and then an AT-determined CNN-based demodulator is activated to recognize the incident OAM modes. Sufficient numeral simulations compare the recognition rates of ATDM-CNN and S-CNN. The results indicate a tremendous improvement owing to the ATDM-CNN demodulator. Base on the ATDM-CNN's significant advantage in OAM recognition, the significant optimization of image data quality is possible in the further correction. As an option, the residual information errors are corrected by jointly using the rank-order adaptive median filter (RAMF) and the very-deep super-resolution (VDSR) network with minor information loss in severe ATs. The data increase resulting from RAMF-VDSR is tested. In conclusion, the proposed two-step system can provide a much higher quality of receiving image data in the OAM-SK-FSO link.

Manifold path selection steering law for pyramid-type single-gimbal control moment gyros

The properties of singularities for pyramid-type single-gimbal control moment gyro (SGCMG) configuration are investigated in this paper. Manifold theory is applied to the SGCMG system and singular surfaces are classified by passability and directionality. A terminal domain is introduced as a small angular momentum space wrapped by internal elliptic singular surfaces and intersected hyperbolic singular surfaces. The hyperbolic singular surfaces in a terminal domain can provide the SGCMG system with a chance to select a manifold path containing no internal elliptic singularity. Thereafter, a manifold path selection (MPS) steering law based on a modified gradient method is proposed, which can achieve internal elliptic singularity avoidance without introducing torque errors. Compared with the other two steering laws, the effectiveness of the proposed method is demonstrated using attitude maneuver examples.

Design method toward creating a three-dimensional asymmetric compound parabolic concentrator

Ideal three-dimensional concentration has been sought for decades in the solar and optical worlds. Nonimaging optics has been able to achieve this thermodynamic maximum concentration limit with certain symmetrical designs, but three-dimensional concentration in which acceptance angles are not symmetric has been a challenge for researchers. Our study outlines a new design approach that hopes to move toward the development of an ideal 3D asymmetric concentrator based on the current asymmetric compound parabolic concentrator (ACPC) design principles. A geometric scheme to revolve a variable ACPC curve about a central axis is presented. The 3D ACPC is compared with a standard compound parabolic concentrator (CPC) in a tilted and angular truncated case. Each design was characterized for performance by examining the angular acceptance/illumination region in angular direction cosine space. Ideal performance was demonstrated for the 3D ACPC design for an off axis circular Lambertian acceptance region in direction cosine plots, whereas the CPC designs failed to maintain a circular Lambertian distribution. High-aperture tilt angles demonstrated an interesting “flaring” of the aperture, which caused a decline in ideal performance. Despite design deformations at high-aperture angles, the 3D ACPC design method presented here is a step toward ideal 3D asymmetric concentration.

Mechanism and method for the full-scale quantitative diagnosis of ball bearings with an inner race fault

A full-scale quantitative diagnostic method is proposed in this study for inner-race fault-size estimation of ball bearings. The fault size is generally estimated using time interval between double impacts if the fault size is smaller than ball angular spacing. This method may be inapplicable when fault sizes become larger than integral multiples of ball angular spacing, which means they have the same time intervals. The analysis of dynamic modelling reveals the existence of a zone where load distribution is unaffected by the inner-race fault and the range of the zone varies monotonously with the fault size. The monotonic relationship between the proportion of fault-free components in axis-orbit time history and fault size can be explored using this mechanism. Hence, a novel evaluation indicator called the axis-orbit deterioration degree (AODD) is proposed in this study. The expression of this indicator is derived from theoretical analysis and signal processing methods. AODD in the simulation and experimental studies can be used to distinguish between faults that differ in size by integral multiples of the ball angular spacing. A new idea is proposed for the quantitative diagnosis method of ball bearings to improve the accuracy of fault size estimation.

Gait event detection based on inter-joint coordination using only angular information

The detection of gait events with wearable sensors is necessary for a robotic system interacting with walking people. Conventional gait phase detection methods are based on machine learning. However, this method cannot detect a gait event every gait cycle because it is difficult to extract characteristic points. Additionally, using only angular information for detection is beneficial because angular information is needed for the control and evaluation of the robots. This paper proposes a novel algorithm for the detection of heel contact and toe-off using the inter-joint coordination of the hip, knee, and ankle joints that has a lower-dimensional structure. The proposed algorithm derives the four planes in the angular space and finds the switching points of the planes. Seven participants walked on force plates that measured the force of the foot against the floor. The error was less than 0.035 s when the gait events were detected after calculating planes using the first gait datum. The change in the patterns of the inter-joint coordination reflected the change in gait phases. Although the data were calculated offline, the results show that the heel contact and toe-off could be detected as soon as the angles were sensed once the planes were derived.

Multidimensional Feature Representation and Learning for Robust Hand-Gesture Recognition on Commercial Millimeter-Wave Radar

This article presents a robust hand-gesture recognition method via multidimensional feature representation and learning specifically designed for commercial frequency-modulated continuous wave (FMCW) multi-input multi-output (MIMO) millimeter-wave radar. First, the optimal configuration of the radar system parameters for the hand-gesture recognition scenario is investigated and a standard procedure to determine the system configuration is given. Then a moving scattering center model is proposed to represent the 3-D point cloud in the range–Doppler (RD)-angular multidimensional feature space. A scattering point detection and tracking algorithm is presented based on a set of motion constraints in terms of position, velocity, and acceleration. It is derived from the space-time continuity of a nonrigid target. Finally, a lightweight multichannel convolutional neural network (CNN) is designed to learn and classify multidimensional gesture features including radial RD and tangential azimuth–elevation. Extensive experiments are carried out with the developed system and a large data set is obtained to train and test the classifier. The results show that the proposed gesture recognition method can effectively distinguish gestures that are easily confused in the RD domain and achieve robust performances under various conditions.

High-accuracy atmospheric turbulence compensation based on a Wirtinger flow algorithm in an orbital angular momentum-free space optical communication system

In this study, high-accuracy atmospheric turbulence compensation based on a Wirtinger flow (WF) algorithm in an orbital angular momentum-free space optical system is demonstrated. An adaptive optics compensation approach based on the WF sensing part is introduced. In addition, the configuration complexity of the wavefront sensing part is reduced to one input. The WF algorithm can be considered as a gradient descent method with two steps, initialisation and iteration, which effectively prevent a fall into the local optimal solution. The compensation performance of the WF algorithm is numerically simulated. The simulation results demonstrate that the compensation can effectively eliminate the impact of atmospheric turbulence. The performance parameters, namely, root-mean-square (RMS) error, mode purity, crosstalk and bit error rate, are significantly improved. In particular, the RMS error reduces to 10−15 compared with 10−2 for the conventional Gerchberg–Saxton algorithm. The performance of the simulations indicates the high effectiveness and accuracy of the WF algorithm.

Unstable Invasion of Sedimenting Granular Suspensions.

We investigate the development of mobility inversion and fingering when a granular suspension is injected radially between horizontal parallel plates of a cell filled with a miscible fluid. While the suspension spreads uniformly when the suspension and the displaced fluid densities are exactly matched, even a small density difference is found to result in a dense granular front which develops fingers with angular spacing that increase with granular volume fraction and decrease with injection rate. We show that the timescale over which the instability develops is given by the volume fraction dependent settling timescale of the grains in the cell. We then show that the mobility inversion and the nonequilibrium Korteweg surface tension due to granular volume fraction gradients determine the number of fingers at the onset of the instability in these miscible suspensions.

Conjugate numerical analysis on bundle effects for integrated heat transfer of LNG ambient air vaporizer

Ambient air vaporizer (AAV) is widely used to vaporize liquefied natural gas (LNG) in gas terminal stations due to its low cost in operating and eco-friendly advantages. Its vaporization efficiency depends on the overall heat transfer performance of the finned tube bundle, which was mostly represented by single tube in previous researches. This paper presented a numerical investigation on the integrated heat transfer of AAV tube bundle considering the heat transfer conjugate problem between different zones. An experimental test combined with the nondimensional correlation in finite space was conducted to validate the numerical model, and the results are in good agreement. The results indicate that the surrounding tubes weakens the field synergy of the flow and heat transfer in inner area of the AAV array, resulting in the decline of heat transfer coefficient from far to near the center of the array. For each finned tube in the AAV bundle, the circumferential local convective heat transfer coefficients decrease from 45° to 225° angular space and then arise from 225° to 360° angular space. Besides, poorer convective heat transfer of the air will lead to the delay of LNG multiphase flow with higher heat transfer intensity due to the conjugate heat transfer in AAV.

A System-Level Engineering Approach for Preliminary Performance Analysis and Design of Global Navigation Satellite System Constellations

This paper presents a system-level engineering approach for the preliminary coverage performance analysis and the design of a generic Global Navigation Satellite System (GNSS) constellation. This analysis accounts for both the coverage requirements and the robustness to transient or catastrophic failures of the constellation. The European GNSS, Galileo, is used as reference case to prove the effectiveness of the proposed tool. This software suite, named GNSS Coverage Analysis Tool (G-CAT), requires as input the state vector of each satellite of the constellation and provides the performance of the GNSS constellation in terms of coverage. The tool offers an orbit propagator, an attitude propagator, an algorithm to identify the visibility region on the Earth's surface from each satellite, and a counter function to compute how many satellites are in view from given locations on the Earth's surface. Thanks to its low computational burden, the tool can be adopted to compute the optimal number of satellites per each orbital plane by verifying if the coverage and accuracy requirements are fulfilled under the assumption of uniform in-plane angular spacing between coplanar satellites.