arXiv subjects

Yunpeng Jia

Publications and source records attributed to Yunpeng Jia.

12 recordsLinked to original sources

arXiv2026

Structure-Semantic Co-optimized Latent Diffusion Model for Fast Visual Anagram Synthesis

Visual anagram is an intriguing form of art creation wherein a single image presents different conceptual interpretations under transformations such as flipping or rotation. Recent work has achieved visual anagram synthesis by leveraging pretrained text-to-image (T2I) diffusion models, yet still suffers from several key limitations including computational inefficiency, suboptimal aesthetic quality, and weak semantic fidelity and expressiveness. This work focuses on generating visual anagrams with substantially improved visual quality at minimal computational cost, thereby advancing intelligent creation of illusionary digital art. To increase image resolution while reducing time overhead, we adapt the cutting-edge parallel denoising algorithm from pixel-based T2I model to the adversarially distilled latent-based one, and accordingly propose a structure-semantic co-optimization (S2CO) framework to counteract the consequent visual degradation. As the core of our approach, S2CO framework comprises three key innovations: (\romannumeral1) null-text structure alignment optimization; (\romannumeral2) semantic enhancement optimization; (\romannumeral3) attention-guided noise fusion. Building upon these components, our method dubbed \textbf{S2CO-Anagram} is able to generate higher-resolution anagram images with noticeably superior visual harmony and semantic faithfulness than related SOTA approaches, all while achieving substantially faster inference speed. Code will be publicly available.

cs.CV↗

arXiv2026

FBSDiff++: Improved Frequency Band Substitution of Diffusion Features for Efficient and Highly Controllable Text-Driven Image-to-Image Translation

With large-scale text-to-image (T2I) diffusion models achieving significant advancements in open-domain image creation, increasing attention has been focused on their natural extension to the realm of text-driven image-to-image (I2I) translation, where a source image acts as visual guidance to the generated image in addition to the textual guidance provided by the text prompt. We propose FBSDiff, a novel framework adapting off-the-shelf T2I diffusion model into the I2I paradigm from a fresh frequency-domain perspective. Through dynamic frequency band substitution of diffusion features, FBSDiff realizes versatile and highly controllable text-driven I2I in a plug-and-play manner (without need for model training, fine-tuning, or online optimization), allowing appearance-guided, layout-guided, and contour-guided I2I translation by progressively substituting low-frequency band, mid-frequency band, and high-frequency band of latent diffusion features, respectively. In addition, FBSDiff flexibly enables continuous control over I2I correlation intensity simply by tuning the bandwidth of the substituted frequency band. To further promote image translation efficiency, flexibility, and functionality, we propose FBSDiff++ which improves upon FBSDiff mainly in three aspects: (1) accelerate inference speed by a large margin (8.9$\times$ speedup in inference) with refined model architecture; (2) improve the Frequency Band Substitution module to allow for input source images of arbitrary resolution and aspect ratio; (3) extend model functionality to enable localized image manipulation and style-specific content creation with only subtle adjustments to the core method. Extensive qualitative and quantitative experiments verify superiority of FBSDiff++ in I2I translation visual quality, efficiency, versatility, and controllability compared to related advanced approaches.

cs.CV↗

arXiv2026

PolGS++: Physically-Guided Polarimetric Gaussian Splatting for Fast Reflective Surface Reconstruction

Accurate reconstruction of reflective surfaces remains a fundamental challenge in computer vision, with broad applications in real-time virtual reality and digital content creation. Although 3D Gaussian Splatting (3DGS) enables efficient novel-view rendering with explicit representations, its performance on reflective surfaces still lags behind implicit neural methods, especially in recovering fine geometry and surface normals. To address this gap, we propose PolGS++, a physically-guided polarimetric Gaussian Splatting framework for fast reflective surface reconstruction. Specifically, we integrate a polarized BRDF (pBRDF) model into 3DGS to explicitly decouple diffuse and specular components, providing physically grounded reflectance modeling and stronger geometric cues for reflective surface recovery. Furthermore, we introduce a depth-guided visibility mask acquisition mechanism that enables angle-of-polarization (AoP)-based tangent-space consistency constraints in Gaussian Splatting without costly ray-tracing intersections. This physically guided design improves reconstruction quality and efficiency, requiring only about 10 minutes of training. Extensive experiments on both synthetic and real-world datasets validate the effectiveness of our method.

cs.CV↗

arXiv2025

PolGS: Polarimetric Gaussian Splatting for Fast Reflective Surface Reconstruction

Efficient shape reconstruction for surfaces with complex reflectance properties is crucial for real-time virtual reality. While 3D Gaussian Splatting (3DGS)-based methods offer fast novel view rendering by leveraging their explicit surface representation, their reconstruction quality lags behind that of implicit neural representations, particularly in the case of recovering surfaces with complex reflective reflectance. To address these problems, we propose PolGS, a Polarimetric Gaussian Splatting model allowing fast reflective surface reconstruction in 10 minutes. By integrating polarimetric constraints into the 3DGS framework, PolGS effectively separates specular and diffuse components, enhancing reconstruction quality for challenging reflective materials. Experimental results on the synthetic and real-world dataset validate the effectiveness of our method.

cs.CV↗

arXiv2025

Offset geometry for extended field-of-view in multi-contrast and multi-scale X-ray microtomography of lung cancer lobectomy specimens

X-ray microtomography is a powerful non-destructive technique allowing 3D virtual histology of resected human tissue. The achievable imaging field-of-view, is however limited by the fixed number of detector elements, enforcing the requirement to sacrifice spatial resolution in order to image larger samples. In applications such as soft-tissue imaging, phase-contrast methods are often employed to enhance image contrast. Some of these methods, especially those suited to laboratory sources, rely on optical elements, the dimensions of which can impose a further limitation on the field-of-view. We describe an efficient method to double the maximum field-of-view of a cone-beam X-ray microtomography system, without sacrificing on spatial resolution, and including multi-contrast capabilities. We demonstrate an experimental realisation of the method, achieving exemplary reconstructions of a resected human lung sample, with a cubic voxel of 10.5 $μ$m linear dimensions, across a horizontal field-of-view of 4.3 cm. The same concepts are applied to free-space propagation imaging of a 2.7 mm segment of the same sample, achieving a cubic voxel of 450 nm linear dimensions. We show that the methodology can be applied at a range of different length-scales and geometries, and that it is directly compatible with complementary implementations of X-ray phase-contrast imaging.

physics.med-ph↗

arXiv2024

Investigating the vertical distribution of the disk as a function of radial action: Results from simulations

Previous research has established a relationship between radial action and scale height in Galactic disks, unveiling a correlation between radial and vertical heating. This finding poses a challenge to our existing comprehension of heating theories and consequently encodes crucial insights into the formation and heating history of Galactic disks. In this study, we perform N-body simulations with the aim of verifying the existence of this correlation between radial action and scale height, thereby enhancing our comprehension of the heating history of Galactic disks. We find that the relationship between radial action and scale height in our simulations can be described by the same functional form observed in previous work. Furthermore, the relationships derived from our simulations align well with those of the Galactic thin disk. However, they do not coincide with the inner thick disk but exhibit a rough correspondence with the outer thick disk, suggesting the possibility that additional heating mechanisms may be required to explain the inner thick disk. We also find that the mean radial action and scale height undergo rapid increases during the initial stages of the simulation, yet remain relatively unchanged as the disk evolves further. By tracing example particles, we uncover a correlation between radial and vertical heating in our simulation: as a particle in the disk gains or loses radial action, its vertical motion tends to oscillate on a more or less extended orbit, accompanied by a tendency to migrate outward or inward, respectively. The massive, long-lasting particles in our simulation contribute to disk heating by solely enhancing the rate of increase in scale height with radial action, while maintaining the functional form that describes the relationship between these two variables.

astro-ph.GA↗

arXiv2024

NeRSP: Neural 3D Reconstruction for Reflective Objects with Sparse Polarized Images

We present NeRSP, a Neural 3D reconstruction technique for Reflective surfaces with Sparse Polarized images. Reflective surface reconstruction is extremely challenging as specular reflections are view-dependent and thus violate the multiview consistency for multiview stereo. On the other hand, sparse image inputs, as a practical capture setting, commonly cause incomplete or distorted results due to the lack of correspondence matching. This paper jointly handles the challenges from sparse inputs and reflective surfaces by leveraging polarized images. We derive photometric and geometric cues from the polarimetric image formation model and multiview azimuth consistency, which jointly optimize the surface geometry modeled via implicit neural representation. Based on the experiments on our synthetic and real datasets, we achieve the state-of-the-art surface reconstruction results with only 6 views as input.

cs.CV↗

arXiv2022

Investigating the vertical distribution of the disk as a function of radial action

As heating processes can broaden the distributions of radial actions and the vertical distributions of the Galactic disks, we investigate the vertical distribution of the Galactic disks as a function of radial action based on Apache Point Observatory Galactic Evolution Experiment(APOGEE) and Gaia data in order to deepen our understanding of the formation and heating history of the Galactic disks. We find that the distributions of radial action for both the thin and thick disks can be approximately described by pseudo-isothermal distributions, which give a statistical measurement for the temperature of the disk as indicated by the mean radial action of the star sample. Estimations of the scale heights in different radial action ranges for these pseudo-isothermal distributions of the disks seem to show fixed relationships between radial action $J_R$ and scale height $h$. We describe these relationships with a two-parameter function of $h = \sqrt{J_R /a} + b$, where $a$ and $b$ are free parameters. When testing with a three-parameter function of $h=\sqrt[α]{J_R/a}+b$, we find that this two-parameter function describes the thin disk well, but we note the function should be used with care for the thick disk. When comparing the best-fit relationships between the inner and outer disk for both of the thin and thick disks, we find that the relationships are nearly the same for the thin disks but are different for the thick disks. The inner thick disk shows a nearly flattened relationship, while the outer thick disk presents a gradually increasing relationship. This work highlights an alternative way to unveil the heating history of the disks by investigating the relationship between scale height and radial action, as these relationships encode the formation and heating history of the Galactic disks.

astro-ph.GA↗

arXiv2018

On the radial metallicity gradient and radial migration effect of the Galactic disk

We study the radial metallicity gradient $Δ[M/H]/ΔR_g$ as a function of [Mg/Fe] and $|Z|$ with the help of a guiding radius based on the Apache Point Observatory Galactic Evolution Experiment and Gaia and then analyze the radial migration effect on the radial metallicity gradient and metallicity-rotation gradient between the Galactic thin and thick disks. The derived trend of gradient $Δ[M/H]/ΔR_g$ versus [Mg/Fe] shows a transition at [Mg/Fe] $\sim 0.18$ dex, below which the gradient is negative and varies a little as [Mg/Fe] increases; however, it changes sharply in [Mg/Fe] ranges of 0.16-0.18, above which the gradient increases linearly with increasing [Mg/Fe], being a positive value at [Mg/Fe]$\gtrsim 0.22$ dex. These positive gradients in the high-[Mg/Fe] populations are found at $|Z| < 0.8$ kpc, and there are nearly no gradients toward higher $|Z|$. By comparing the metallicity distributions, the radial metallicity gradients $Δ[M/H]/ΔR$ and the metallicity-rotation gradients between the total sample and $|R-R_g|<2$ kpc subsample (or $|R-R_g|>2$ kpc subsample), we find that, for the thick disk, blurring flattens the gradient $Δ[M/H]/ΔR$ and favors metal-poor high-eccentricity stars. These stars are responsible for the measured positive metallicity-rotation gradient of the thick disk.

astro-ph.GA↗

arXiv2016

Kinematics of the Galactic disk from LAMOST Dwarf sample

Based on the LAMOST survey and Sloan Digital Sky Survey (SDSS), we use low-resolution spectra of 130,043 F/G-type dwarf stars to study the kinematics and metallicity properties of the Galactic disk. Our study shows that the stars with poorer metallicity and larger vertical distance from Galactic plane tend to have larger eccentricity and velocity dispersion. After separating the sample stars into likely thin-disk and thick-disk sub-sample, we find that there exits a negative gradient of rotation velocity $V_ϕ$ with metallicity [Fe/H] for the likely thin-disk sub-sample, and the thick-disk sub-sample exhibit a larger positive gradient of rotation velocity with metallicity. By comparing with model prediction, we consider the radial migration of stars appears to have influenced on the thin-disk formation. In addition, our results shows that the observed thick-disk stellar orbital eccentricity distribution peaks at low eccentricity ($e \sim 0.2$) and extends to a high eccentricity ($e \sim 0.8$). We compare this result with four thick-disk formation simulated models, and it imply that our result is consistent with gas-rich merger model.

astro-ph.GA↗

arXiv2015

Photometric Metallicity Calibration with SDSS and SCUSS and its Application to distant stars in the South Galactic Cap

Based on SDSS g, r and SCUSS (South Galactic Cap of u-band Sky Survey) $u$ photometry, we develop a photometric calibration for estimating the stellar metallicity from $u-g$ and $g-r$ colors by using the SDSS spectra of 32,542 F- and G-type main sequence stars, which cover almost $3700$ deg$^{2}$ in the south Galactic cap. The rms scatter of the photometric metallicity residuals relative to spectrum-based metallicity is $0.14$ dex when $g-r<0.4$, and $0.16$ dex when $g-r>0.4$. Due to the deeper and more accurate magnitude of SCUSS $u$ band, the estimate can be used up to the faint magnitude of $g=21$. This application range of photometric metallicity calibration is wide enough so that it can be used to study metallicity distribution of distant stars. In this study, we select the Sagittarius (Sgr) stream and its neighboring field halo stars in south Galactic cap to study their metallicity distribution. We find that the Sgr stream at the cylindrical Galactocentric coordinate of $R\sim 19$ kpc, $\left| z\right| \sim 14$ kpc exhibits a relative rich metallicity distribution, and the neighboring field halo stars in our studied fields can be modeled by two-Gaussian model, with peaks respectively at [Fe/H]$=-1.9$ and [Fe/H]$=-1.5$.

astro-ph.GA↗

arXiv2014

Estimation of Absolute Magnitude-dependent Galactic Model Parameters In Intermediate Latitude With SDSS and SCUSS

Based on SDSS and South Galactic Cap of u-band Sky Survey (SCUSS) early data, we use star counts method to estimate the Galactic structure parameters in an intermediate latitude with 10,180 main-sequence (MS) stars in absolute magnitude interval of $4 \leq M_r \leq 13$. We divide the absolute magnitude into five intervals:$4 \leq M_r < 5$, $5 \leq M_r < 6$, $6 \leq M_r < 8$, $8 \leq M_r < 10$, $10 \leq M_r \leq 13$, and estimate the Galactic structure parameters in each absolute magnitude interval to explore their possible variation with the absolute magnitude. Our study shows the parameters depend on absolute magnitude. For the thin disk, the intrinsic faint MS stars have large local space density and they tend to stay close to the Galactic plane. A plausible explanation is that faint MS stars with long lifetime experience long gravitational interaction time result in a short scaleheight. However, for the thick disk, the parameters show a complex trend with absolute magnitude, which may imply the complicated original of the thick disk. For the halo, the intrinsic faint MS stars have large local density and small axial ratio, which indicate a flattened inner halo and a more spherical outer halo.

astro-ph.GA↗