Horizontal Stripes Research Articles

Margin-enhanced average precision optimization for visible-infrared person re-identification

Average precision-based loss functions are effective for training deep neural networks in ranking tasks and single-modality person re-identification. However, they have difficulty in uniformly ranking cross-modality samples due to high intra-modality similarity. To address this limitation, this paper proposes a Margin-Enhanced Average Precision (MEAP) optimization approach for cross-modality person re-identification. MEAP integrates average precision optimization with inter-class and cross-modality margin parameters. The inter-class margin improves performance by increasing class separation, while the cross-modality margin enhances performance by prioritizing cross-modality positive samples. Additionally, we propose an innovative algorithm to augment visible and infrared person images using horizontal stripes, aiming to bridge the gap between the two modalities by creating diverse and enriched training data. Experiments on two public datasets demonstrate the effectiveness of our approach in comparison to state-of-the-art methods. MEAP and horizontal stripe augmentation significantly improve accuracy and robustness in matching individuals across different modalities. The code is available at: https://github.com/NihatTekeli/meapnet

Preponderance of triaxial shapes in atomic nuclei predicted by the proxy-SU(3) symmetry

Abstract The proxy-SU(3) symmetry predicts, in a parameter-free way, based only on the Pauli principle and the short-range nature of the nucleon-nucleon interaction, non-vanishing values of the collective variable gamma almost everywhere across the nuclear chart. Substantial triaxiality with gamma between 15 and 45 degrees is proved to be expected along horizontal and vertical stripes on the nuclear chart, covering the nucleon numbers 22-26, 34-48, 74-80, 116-124, 172-182. Empirical support for these stripes is found by collecting all even-even nuclei for which the first two excited 2+ states are known, along with the B(E2)s connecting them, as well as the second 2+ state to the ground state. The stripes are related to regions in which oblate SU(3) irreducible representations appear, bearing similarity to the appearance of triaxiality within the SU(3)* dynamical symmetry of the interacting boson model-2. Detailed comparisons of the proxy-SU(3) predictions to the data and to predictions by state-of-the-art Monte Carlo shell model calculations for deformed N=94, 96, 98 isotones in the rare earth region show good overall agreement, with the exception of Z=70 and N=94, which correspond to fully symmetric proxy-SU(3) irreps, suggesting that the latter are an artifact of the method which can be amended by considering the influence of the neighboring irreps.

Relative Radiometric Correction Method Based on Temperature Normalization for Jilin1-KF02

The optical remote sensors carried by the Jilin-1 KF02 series satellites have an imaging resolution better than 0.5 m and a width of 150 km. There are radiometric problems, such as stripe noise, vignetting, and inter-slice chromatic aberration, in their raw images. In this paper, a relative radiometric correction method based on temperature normalization is proposed for the response characteristics of sensors and the structural characteristics of optical splicing of Jilin-1 KF02 series satellites cameras. Firstly, a model of temperature effect on sensor output is established to correct the variation of sensor response output digital number (DN) caused by temperature variation during imaging process, and the image is normalized to a uniform temperature reference. Then, the horizontal stripe noise of the image is eliminated by using the sensor scan line and dark pixel information, and the vertical stripe noise of the image is eliminated by using the method of on-orbit histogram statistics. Finally, the method of superposition compensation is used to correct the vignetting area at the edge of the image due to the lack of energy information received by the sensor so as to ensure the consistency of the image in color and image quality. The proposed method is verified by Jilin-1 KF02A on-orbit images. Experimental results show that the image response is uniform, the color is consistent, the average Streak Metrics (SM) is better than 0.1%, Root-Mean-Square Deviation of the Mean Line (RA) and Generalized Noise (GN) are better than 2%, Relative Average Spectral Error (RASE) and Relative Average Spectral Error (ERGAS) are greatly improved, which are better than 5% and 13, respectively, and the relative radiation quality is obviously improved after relative radiation correction.

Mitigation of DMM-induced stripe patterns in synchrotron X-ray radiography through dynamic tilting.

In synchrotron X-ray radiography, achieving high image resolution and an optimal signal-to-noise ratio (SNR) is crucial for the subsequent accurate image analysis. Traditional methods often struggle to balance these two parameters, especially in situ applications where rapid data acquisition is essential to capture specific dynamic processes. For quantitative image data analysis, using monochromatic X-rays is essential. A double multilayer monochromator (DMM) is successfully used for this aim at the BAMline, BESSY II (Helmholtz Zentrum Berlin, Germany). However, such DMMs are prone to producing an unstable horizontal stripe pattern. Such an unstable pattern renders proper signal normalization difficult and thereby causes a reduction of the SNR. We introduce a novel approach to enhance SNR while preserving resolution: dynamic tilting of the DMM. By adjusting the orientation of the DMM during the acquisition of radiographic projections, we optimize the X-ray imaging quality, thereby enhancing the SNR. The corresponding shift of the projection during this movement is corrected in post-processing. The latter correction allows a good resolution to be preserved. This dynamic tilting technique enables the homogenization of the beam profile and thereby effectively reduces noise while maintaining high resolution. We demonstrate that data captured using this proposed technique can be seamlessly integrated into the existing radiographic data workflow, as it does not need hardware modifications to classical X-ray imaging beamline setups. This facilitates further image analysis and processing using established methods.

Learning differentiable categorical regions with Gumbel-Softmax for person re-identification

Locating diverse body parts and perceiving part visibility are essential to person re-identification (re-ID). Most existing methods employ an extra model, e.g., pose estimation or human parsing, to locate parts, or generate pseudo labels to train the part locator incorporated with the re-ID model. In this paper, we aim at learning diverse horizontal stripes with foreground refinement to pursue pixel-level part alignment via only using person identity labels. Specifically, we proposed a Gumbel-Softmax based Differential Categorical Region (DCR) learning method and make two contributions. (1) A stripe-wise regularization. Given an image, the part locator produce part probability maps. The continuous values in the probability maps are discretized into zero or argmax value in the horizontal stripes by the Gumbel-Softmax. Gumbel-Softmax allows us to use the argmax discrete value for part diversity regularization in the forward pass, but can still estimate gradients in the backward pass. (2) A self-refinement method to suppress the background noise in the stripes. We employ a lightweight foreground perception head to produce foreground probability map with only person identity labels supervision. Benefits from discretization of the categorical stripes, we can conveniently obtain the part pseudo label by element-wise multiplying the categorical stripes with foreground probability map. Finally, DCR can locate the body parts at pixel-level and extract part-aligned representation. Experimental results on both holistic and occluded re-ID datasets confirm that our approach significantly improves the learned representation and the achieved performance is on par with the state-of-the-art methods. The code is available at https://github.com/deepalchemist/differentiable-categorical-region

An efficient parallel self-attention transformer for CSI feedback

In massive multi-input multi-output (MIMO) systems, it is necessary for user equipment (UE) to transmit downlink channel state information (CSI) back to the base station (BS). As the number of antennas increases, the feedback overhead of CSI consumes a significant amount of uplink bandwidth resources. To minimize the bandwidth overhead, we propose an efficient parallel attention transformer, called EPAformer, a lightweight network that utilizes the transformer architecture and efficient parallel self-attention (EPSA) for CSI feedback tasks. The EPSA expands the attention area of each token within the transformer block effectively by dividing multiple heads into parallel groups and conducting self-attention in horizontal and vertical stripes. The proposed EPSA achieves better feature compression and reconstruction. The simulation results display that the EPAformer surpasses previous deep learning-based approaches in terms of reconstruction performance and complexity.

CSWin-UNet: Transformer UNet with cross-shaped windows for medical image segmentation

Deep learning, especially convolutional neural networks (CNNs) and Transformer architectures, have become the focus of extensive research in medical image segmentation, achieving impressive results. However, CNNs come with inductive biases that limit their effectiveness in more complex, varied segmentation scenarios. Conversely, while Transformer-based methods excel at capturing global and long-range semantic details, they suffer from high computational demands. In this study, we propose CSWin-UNet, a novel U-shaped segmentation method that incorporates the CSWin self-attention mechanism into the UNet to facilitate horizontal and vertical stripes self-attention. This method significantly enhances both computational efficiency and receptive field interactions. Additionally, our innovative decoder utilizes a content-aware reassembly operator that strategically reassembles features, guided by predicted kernels, for precise image resolution restoration. Our extensive empirical evaluations on diverse datasets, including synapse multi-organ CT, cardiac MRI, and skin lesions demonstrate that CSWin-UNet maintains low model complexity while delivering high segmentation accuracy.

Edge-enhanced infrared image super-resolution reconstruction model under transformer

Infrared images have important applications in military, security and surveillance fields. However, limited by technical factors, the resolution of infrared images is generally low, which seriously limits the application and development of infrared images in various fields. To address the problem of difficult recovery of edge information and easy ringing effect in the super-resolution reconstruction process of infrared images, an edge-enhanced infrared image super-resolution reconstruction model TESR under transformer is proposed. The main structure of this model is transformer. First, in view of the problem of difficult recovery of edge information of infrared images, an edge detection auxiliary network is designed, which can obtain more accurate edge information from the input low-resolution images and enhance the edge details during image reconstruction; then, the CSWin Transformer is introduced to compute the self-attention of horizontal and vertical stripes in parallel, so as to increase the receptive field of the model and enable it to utilize features with higher semantic levels. The super-resolution reconstruction model proposed in this paper can extract more comprehensive image information, and at the same time, it can obtain more accurate edge information to enhance the texture details of super-resolution images, and achieve better reconstruction results.

Traffic safety comprehensive evaluation of urban tunnel visual guiding system based on extension matter-element model: a case study in tunnel curves

Objective The visual guiding system, as a tunnel traffic safety improvement method by using visual guiding facilities to actively guide driving safely, has been widely used in countries with many tunnels, in recent years. This paper aims to quantitatively study the comprehensive evaluation of traffic safety of the visual guiding system in tunnels, which has certain engineering application value and can provide support for the quantitative evaluation and optimal design of tunnel traffic safety. Methods Based on the analysis of the relevant factors of urban tunnel traffic safety, a multi-factor comprehensive evaluation system with 5 upper-level indicators and 12 basic-level indicators was proposed. Considering the independent and incompatible indicators, a comprehensive evaluation method of traffic safety of the visual guiding system in urban tunnels was constructed by using the extension matter-element model. Taking the scene of 4 types of tunnel curves, such as no facilities, horizontal stripe, chevron alignment sign, and LED arch, as examples, the comprehensive evaluation of various schemes were carried out by using simulation tests. Results The traffic safety comprehensive evaluation system of visual guiding system in urban tunnels can be analyzed from five aspects: perception reaction, guidance ability, driver factor, driving task, and facility appearance. The results demonstrated significant the comprehensive evaluation result of the target level of scene 1 was L4, scene 2 was L3, scene 3 was L2, and scene 4 was L1. That is, the final results of the comprehensive evaluation of the four scenes were poor, medium, good, and very good, respectively. Conclusions In the scheme of visual guiding system for urban tunnel curves, the effectiveness of the three types of designs, from high to low, was the LED arch, chevron alignment sign, and horizontal stripe, and the safety of the scene without facilities was the lowest. Hence, setting the LED arch in the urban tunnel curve has a good effect in the aspects of guidance ability, sight distance, and sight zone, and is conducive to the driver’s perception reaction and driving task.

Evaluating the impact of bird collision prevention glazing patterns on window views

Optimal design of bird collisions prevention glazing patterns is pivotal in the protection of avian ecology and provision of quality window views for building occupants. However, a research gap currently exists regarding the impact of different pattern designs on the quality of window views. Therefore, this study evaluates the influence of various bird collision prevention pattern designs on window view clarity. A total of 18 patterns, including dot matrix, vertical stripes, and horizontal stripes, were designed and evaluated. The patterns had varying combinations of coverage rates and spacing ratios. Sixty participants were recruited to participate in an experimental survey using virtual reality (VR) technology. They were asked to rate the level of clarity of window views observed through different glazing pattern designs. The results indicated that the bird collision prevention glazing pattern designs significantly influenced occupants’ perception of visual clarity. Designs featuring horizontal and vertical stripes attained higher visual clarity ratings compared to the dot matrix pattern. Additionally, visual clarity significantly decreased with increasing coverage rates and designs with larger spacing ratios were associated with higher clarity than smaller spacing ratios. Most participants were willing to sacrifice window view quality in order to minimize bird fatalities. These findings can guide the development of window view evaluation criteria in existing green building standards, by accounting for the trade-off between avian conservation and window view clarity.

Impact of sidewall information on drivers’ visual load among different lanes: Traditional guiding facilities and decorated pattern

Objectives This research aims to: (i) compare the effects of different sidewall entrainment facilities on drivers’ visual behavior; (ii) compare the effects of the same sight entrainment facilities on drivers in different lanes; (iii) give recommendations for engineering applications based on the results of the study; Methods The study designed four different scenes, each with symmetrically designed visual facilities on the both sidewalls of the tunnel, scene a represents a typical urban tunnel in China (horizontal stripes on sidewalls), scene b includes vertical stripes on sidewalls in addition to scene a, scene c introduces an LED-arch based on scene b, and scene d features a rhythmic pattern (Wave pattern on sidewalls). 30 participants, 21 men and 9 women, aged 21–54, drove the four scenes. Eye movement data of participants in each lane for different scenes were collected using an eye-tracking device. Visual performance indicators including fixation duration, number of fixations, saccade duration, and saccade amplitude were utilized to comprehensively evaluate drivers’ visual behavior. Factor analysis was employed to analyze the impact of different visual guiding facilities on drivers’ visual searching abilities. Results There is a significant effect of sidewall guiding facilities and lane location on drivers’ visual behavior and loading. Across scenes, drivers’ visual load is ranked as follows, from highest to lowest: scene a (baseline) > scene b (horizontal stripes added to scene a) > scene c (LED-arch added to scene b) > scene d (Wave pattern). Furthermore, under the same scene, drivers’ visual load in each lane is ranked in descending order: Middle lane > Right lane > Left lane. Conclusion Due to the effect of the tunnel structure on the drivers’ visual field, drivers in the left lane have the highest visual load in any scenario compared to the other two lanes, which can be ameliorated but not eliminated. Traditional guiding facilities and decorated pattern both improve the visual behavior and reduce drivers’ visual load in urban tunnels, especially in scene c and scene d, but scene d should not be used for the entire length of the tunnel in order to prevent driver distraction. In engineering practice, scene c (LED-arch added to scene b) can be set up in general sections of urban tunnels, and decorated pattern can be added to fatigue reminder regions to alleviate driving fatigue.

Deep-GD: Deep Learning based Automatic Garment Defect Detection and Type Classification

Garment defect detection has been successfully implemented in the quality quick response system for textile manufacturing automation. Defects in the production of textiles waste a lot of resources and reduce the quality of the finished goods. It is challenging to detect garment defects automatically because of the complexity of images and variety of patterns in textiles. This study presented a novel deep learning-based Garment defect detection framework named as Deep-GD model for sequentially identifying image defects in patterned garments and classify the defect types. Initially, the images are gathered from the HKBU database and bilateral filters are used in the pre-processing of images to remove distortions. A squeeze-and-excitation network (SE-net) combined with Random Decision Forest Classifier with Bayesian Optimization (RDF-BO) algorithm is used to detect and classify garment defects. By analyzing the differences among the original and reconstruction images, the unsupervised technique trains to rebuild the fabric pattern. The SE-net is used to identify certain fabric flaws in the garments of the pre-processed images. Then, the defects-related garments are processed using RDF-BO algorithm for classifying the garment defect from these regions. Finally, the proposed Deep-GD model is used for classifying defected fabric into 12 classes such as Defect free, soil stain, oil stain, Double end, Snarls, Miss, Horizontal stripes, Lumpy, Dye spot, Fall out, Hairiness, tiny hole. The proposed Deep-GD model achieves the overall classification accuracy of 97.16%, which is comparatively superior to the existing techniques.

Investigation on the elimination of striations in NDE of glass fiber composites

Glass fiber composites are widely used due to their unique performance advantages. However, defects in glass fiber composites can significantly affect their performance, making nondestructive evaluation necessary. During the X-ray spectral imaging of glass fiber composites, the 0/90° cross-stacking mode of glass fiber cloth results in irregular horizontal and vertical stripe noise, which seriously reduces the clarity of defect detection. Therefore, an appropriate algorithm is required to remove the noise. In this study, by analyzing the Fourier spectrum of X-ray spectral images of glass fiber composites, it was found that the stripe information is concealed in the vertical and horizontal bright lines in the middle of the Fourier spectrum image. A novel cross-sector filter was designed, and a stripe noise removal algorithm based on the cross-sector filter was proposed. The degree of the filter’s central angle can be adjusted according to the distribution of the stripe noise in the frequency domain, which removes the noise without losing much useful image information.

A new snouted treefrog (Anura, Hylidae, Scinax) from fluvial islands of the Juruena River, southern Brazilian Amazonia.

Southern Amazonia is one of the less-explored regions by anuran taxonomists. We describe a small new species of snouted treefrog, genus Scinax, from this region, from a fluvial archipelago in the Juruena River, state of Mato Grosso, Brazil. The description is based on external morphology of adults and tadpoles, advertisement call and molecular data. The species is phylogenetically related to other snouted treefrogs of the Scinax cruentomma species group and shows the most southeastern distribution in Amazonia among its close relatives. It is distinguished from congeners mainly by its larger adult body size and bilobate vocal sac that reaches the level of the pectoral fold, a reddish-brown horizontal stripe on the iris, dark melanophores or blotches on the vocal sac and the throat of females, and the uniformly brown posterior portion of the thigh. The advertisement call comprises one pulsed note emitted at regular intervals, with a duration of 189-227 ms, 30-35 pulses/note and a dominant frequency of 2,250-2,344 Hz. The type locality is suffering several environmental impacts, including illegal mining, overfishing, unsustainable agriculture, uncontrolled logging and degradation associated with the construction of new hydroelectric dams. Further study of the biology and regional distribution of the new species is required to propose mitigation measures needed for its conservation.

Description of a rock-dwelling cichlid that re-invaded the sand substrate in Lake Malai, Africa.

Pseudotropheus Regan had been a catch-all genus for many of the rock-dwelling cichlids of Lake Malai, known as mbuna, for over half a century. Although many of the species previously assigned to this genus have since been allocated to other genera there are still about a dozen species in the genus that do not appear to be closely related to its type species, P. williamsi, but for which no better placement could be found. Pseudotropheus livingstonii (Boulenger) is one such species and the new species described herein appears to be closely related and has for that reason been temporarily assigned to Pseudotropheus. Pseudotropheus likomae n. sp. from Likoma Island can be diagnosed from all other Pseudotropheus spp., by having five or fewer distinct vertical bars below the dorsal fin and two horizontal stripes in mature individuals.

Improved cycling stability and ON/OFF ratio of SrFeOx topological phase transition memristors using a La0.7Sr0.3MnO3 bottom electrode

Single-crystal BM-SFO films with horizontal superlattice-like stripes were epitaxially grown on LSMO bottom electrodes and Au/SFO/LSMO memristors exhibit excellent I–V cycling stability and improved ON/OFF ratio.

ANALYTICAL CALCULATION OF MAGNETOELECTRIC VALVE MOTOR AND DEFINITION INDUCTIVE PARAMETERS OF ITS WINDINGS

To implement the dynamic properties of the designed valve motors and achieve the proper quality of transient processes in the drives of complex mechanisms, it is necessary to study their electromagnetic and functional characteristics with the calculation of the inductive parameters of the windings.
 
 The purpose of the study is to study the electromagnetic and functional characteristics of the magnetoelectric valve motor and determine the inductive parameters of its windings.
 
 Materials and methods. The magnetic field of the machine is divided into a set of horizontal stripes (mediums), analyzed independently of each other. It is assumed that the magnetic field in these stripes is plane-parallel, and at the boundaries of the selected stripes the conjugation conditions are met: scalar magnetic potentials and normal components of magnetic induction do not undergo a jump (discontinuity). The magnetomotive force of the stator winding, the coercive force of magnets and the magnetization of soft magnetic ferromagnetic media are considered as sources of the magnetic field. Expressions for describing scalar magnetic potentials and normal components of magnetic induction in selected bands are obtained using the Fourier variable separation method.
 
 Research results. A study of the electromagnetic and functional characteristics of the motor was carried out using a mathematical model of the magnetoelectric valve motor based on the Fourier method of separation of variables with the division of the active region into homogeneous sections in the form of magnetic sheets. The model enables to determine magnetic inductions for any coordinate points x, y of the active region of the motor magnetic system. Expressions describing the inductive parameters of the windings are obtained, and the equations of the state of the motor are given for the case when the currents in the phase windings of the stator are in antiphase with their no-load EMF.
 
 Conclusions. Mathematical models of valve motors, built on the basis of the Fourier variable separation method, make it possible to determine the electromagnetic characteristics of motors and the inductive parameters of windings necessary for calculating transient processes. The most effective is valve motor control, when the phase currents are in antiphase with the no-load EMF. In this mode, with the absence of longitudinal current and with a power factor close to unity, the equations of state of the motor are of the second order.

Image processing methods based on physical models

In order to achieve better image processing results, the author proposes research on image processing methods based on physical models. The author first developed a simple and user-friendly GUIDE interactive interface based on Matlab software and image processing technology. Calculate and process the interference fringe patterns captured in static and dynamic experimental measurements in optical experiments separately to achieve experimental measurement objectives. Secondly, the image processing program was analyzed to highlight important information using methods such as Gaussian blur and binarization, and algorithms were used to construct surface 3D images or detect stripe motion features, establish 3D models, and accurately calculate corresponding experimental results. The final experimental results indicate that the phase transition value of piezoelectric devices increases linearly with the increase of voltage, which is consistent with the working characteristics of piezoelectric devices in the linear region. This program can analyze the image sets of horizontal, vertical, and oblique stripes, and calculate the piezoelectric moduli of piezoelectric devices to be 3.355 nm/V, 3.120 nm/V, and 3.557 nm/V, respectively, with relative errors of E1 = 4.9 %, E2 = 2.9 %, and E3 = 12 %. This method is suitable for various complex experimental measurement scenarios and provides an intuitive method for innovative design of optical interference experimental measurement teaching.

Learning Semantics-Consistent Stripes With Self-Refinement for Person Re-Identification.

Aligning human parts automatically is one of the most challenging problems for person re-identification (re-ID). Recently, the stripe-based methods, which equally partition the person images into the fixed stripes for aligned representation learning, have achieved great success. However, the stripes with fixed height and position cannot well handle the misalignment problems caused by inaccurate detection and occlusion and may introduce much background noise. In this article, we aim at learning adaptive stripes with foreground refinement to achieve pixel-level part alignment by only using person identity labels for person re-ID and make two contributions. 1) A semantics-consistent stripe learning method (SCS). Given an image, SCS partitions it into adaptive horizontal stripes and each stripe is corresponding to a specific semantic part. Specifically, SCS iterates between two processes: i) clustering the rows to human parts or background to generate the pseudo-part labels of rows and ii) learning a row classifier to partition a person image, which is supervised by the latest pseudo-labels. This iterative scheme guarantees the accuracy of the learned image partition. 2) A self-refinement method (SCS+) to remove the background noise in stripes. We employ the above row classifier to generate the probabilities of pixels belonging to human parts (foreground) or background, which is called the class activation map (CAM). Only the most confident areas from the CAM are assigned with foreground/background labels to guide the human part refinement. Finally, by intersecting the semantics-consistent stripes with the foreground areas, SCS+ locates the human parts at pixel-level, obtaining a more robust part-aligned representation. Extensive experiments validate that SCS+ sets the new state-of-the-art performance on three widely used datasets including Market-1501, DukeMTMC-reID, and CUHK03-NP.

Single-Shot, Pixel-Encoded Strip Patterns for High-Resolution 3D Measurement

In this research, we combined two distinct, structured light methods, the single-shot pseudo-random sequence-based approach and the time-multiplexing stripe indexing method. As a result, the measurement resolution of the single-shot, spatially encoded, pseudo-random sequence-based method improved significantly. Since the time-multiplexed stripe-indexed-based techniques have a higher measurement resolution, we used varying stripes to enhance the measurement resolution of the pseudo-random sequence-based approaches. We suggested a multi-resolution 3D measurement system that consisted of horizontal and vertical stripes with pixel sizes ranging from 8 × 8 to 16 × 16. We used robust pseudo-random sequences (M-arrays) to controllably distribute various stripes in a pattern. Since single-shape primitive characters only contribute one feature point to the projection pattern, we used multiple stripes instead of single-shape primitive symbols. However, numerous stripes will contribute multiple feature points. The single character-based design transforms into an increased featured size pattern when several stripes are employed. Hence, the projection pattern contains a much higher number of feature points. So, we obtained a high-resolution measurement. Each stripe in the captured image is located using adaptive grid adjustment and stripe indexing techniques. The triangulation principle is used to measure 3D.