arXiv subjects

Wenbo Li

Publications and source records attributed to Wenbo Li.

At least 199 records · Page 11Linked to original sources

arXiv2019

Three-dimensional laser printing of macro-scale glass objects at a micro-scale resolution

Three-dimensional (3D) printing has allowed for production of geometrically complex 3D objects with extreme flexibility, which is currently undergoing rapid expansions in terms of materials, functionalities, as well as areas of application. When attempting to print 3D microstructures in glass, femtosecond laser induced chemical etching (FLICE) has proved itself a powerful approach. Here, we demonstrate fabrication of macro-scale 3D glass objects of large heights up to ~3.8 cm with a well-balanced (i.e., lateral vs longitudinal) spatial resolution of ~20 μm. The remarkable accomplishment is achieved by revealing an unexplored regime in the interaction of ultrafast laser pulses with fused silica which results in aberration-free focusing of the laser pulses deeply inside fused silica.

physics.app-ph↗

arXiv2018

Two-scale methods for convex envelopes

We develop two-scale methods for computing the convex envelope of a continuous function over a convex domain in any dimension.This hinges on a fully nonlinear obstacle formulation [A. M. Oberman, "The convex envelope is the solution of a nonlinear obstacle problem", Proc. Amer. Math. Soc. 135(6):1689--1694, 2007]. We prove convergence and error estimates in the max norm. The proof utilizes a discrete comparison principle, a discrete barrier argument to deal with Dirichlet boundary values, and the property of flatness in one direction within the non-contact set. Our error analysis extends to a modified version of the finite difference wide stencil method of [A. M. Oberman, "Computing the convex envelope using a nonlinear partial differential equation", Math. Models Meth. Appl. Sci, 18(05):759--780, 2008].

math.NA↗

arXiv2018

Evolvement Constrained Adversarial Learning for Video Style Transfer

Video style transfer is a useful component for applications such as augmented reality, non-photorealistic rendering, and interactive games. Many existing methods use optical flow to preserve the temporal smoothness of the synthesized video. However, the estimation of optical flow is sensitive to occlusions and rapid motions. Thus, in this work, we introduce a novel evolve-sync loss computed by evolvements to replace optical flow. Using this evolve-sync loss, we build an adversarial learning framework, termed as Video Style Transfer Generative Adversarial Network (VST-GAN), which improves upon the MGAN method for image style transfer for more efficient video style transfer. We perform extensive experimental evaluations of our method and show quantitative and qualitative improvements over the state-of-the-art methods.

cs.CV↗

arXiv2018

Who did What at Where and When: Simultaneous Multi-Person Tracking and Activity Recognition

We present a bootstrapping framework to simultaneously improve multi-person tracking and activity recognition at individual, interaction and social group activity levels. The inference consists of identifying trajectories of all pedestrian actors, individual activities, pairwise interactions, and collective activities, given the observed pedestrian detections. Our method uses a graphical model to represent and solve the joint tracking and recognition problems via multi-stages: (1) activity-aware tracking, (2) joint interaction recognition and occlusion recovery, and (3) collective activity recognition. We solve the where and when problem with visual tracking, as well as the who and what problem with recognition. High-order correlations among the visible and occluded individuals, pairwise interactions, groups, and activities are then solved using a hypergraph formulation within the Bayesian framework. Experiments on several benchmarks show the advantages of our approach over state-of-art methods.

cs.CV↗

arXiv2018

STS Classification with Dual-stream CNN

The structured time series (STS) classification problem requires the modeling of interweaved spatiotemporal dependency. most previous STS classification methods model the spatial and temporal dependencies independently. Due to the complexity of the STS data, we argue that a desirable STS classification method should be a holistic framework that can be made as adaptive and flexible as possible. This motivates us to design a deep neural network with such merits. Inspired by the dual-stream hypothesis in neural science, we propose a novel dual-stream framework for modeling the interweaved spatiotemporal dependency, and develop a convolutional neural network within this framework that aims to achieve high adaptability and flexibility in STS configurations from various diagonals, i.e., sequential order, dependency range and features. The proposed architecture is highly modularized and scalable, making it easy to be adapted to specific tasks. The effectiveness of our model is demonstrated through experiments on synthetic data as well as benchmark datasets for skeleton based activity recognition.

cs.CV↗

arXiv2018

Aberration-free three dimensional micromachining in glass with spatiotemporally shaped femtosecond laser pulses

We observe that focusing a femtosecond laser beam simultaneously chirped in time and space domains in glass can efficiently suppress the optical aberration caused by the refractive index mismatch at the interface of air and the glass sample. We then demonstrate three dimensional microprocessing in glass with a nearly invariant spatial resolution for a large range of penetration depth between 0.25 mm and 9 mm without any aberration correction.

physics.optics↗

arXiv2018

High-throughput high-resolution multi-scale three dimensional laser printing

Throughput and resolution are both of critical importance for modern manufacturing, however, they are usually contradictory with each other. Here, a high-throughput high-resolution three dimensional (3D) printing is demonstrated by incorporating a simultaneous spatiotemporal focusing (SSTF) scheme in two-photon polymerization (TPP). Remarkably, the SSTF can ensure generation of a spherical focal spot of which the size depends linearly on the laser power. Thus, the resolution can be continuously adjusted from sub-10 μm to ~40 μm by only increasing the laser power. A multi-scale 3D structure is fabricated using this technique, for which the regions of coarse and fine feature sizes are produced at high and low resolutions, respectively.

physics.app-ph↗

arXiv2017

Centimeter-scale superfine three-dimensional printing with femtosecond laser two-photon polymerization

Nowadays three-dimensional (3D) printing has been widely used for producing geometrically complex 3D structures from a broad range of materials such as ceramics, metals, polymers, semiconductors, etc. Although it has been demonstrated that a fabrication resolution as high as ~100 nm can be achieved in 3D printing based on two photon polymerization (TPP), the end product size of TPP is typically on millimeter scale limited by the short working distance of high-numerical-aperture focal lens. Here we present a method based on simultaneous spatiotemporal focusing (SSTF) of the femtosecond laser pulses that enables to fabricate centimeter-scale 3D structures of fine features with TPP. We also demonstrate an isotropic spatial resolution which can be continuously tuned in the range of ~10 um and ~40 um by only varying the power of femtosecond laser, making this technique extremely flexible and easy to implement. We fabricate several Chinese guardian lions of a maximum height of 0.6 cm and a Terra Cotta Warrior of a height of 1.3 cm using this method.

physics.app-ph↗

arXiv2016

POI: Multiple Object Tracking with High Performance Detection and Appearance Feature

Detection and learning based appearance feature play the central role in data association based multiple object tracking (MOT), but most recent MOT works usually ignore them and only focus on the hand-crafted feature and association algorithms. In this paper, we explore the high-performance detection and deep learning based appearance feature, and show that they lead to significantly better MOT results in both online and offline setting. We make our detection and appearance feature publicly available. In the following part, we first summarize the detection and appearance feature, and then introduce our tracker named Person of Interest (POI), which has both online and offline version.

cs.CV↗

arXiv2010

Time Variable Cosmological Constants from Cosmological Horizons

In this paper, motivated from the fact that a de Sitter cosmological boundary corresponds to a positive cosmological constant, we consider time variable cosmological constants, dubbed {\it horizon cosmological constants}. The horizon cosmological constants correspond to Hubble horizon, future event horizon and particle horizon are discussed respectively. When the Hubble horizon is taken as a cosmological length scale, the effective equation of state of horizon cosmological constant is quintessence-like. The values of model parameter $c$ will determine the current status of our universe. When particle horizon is taken as the cosmological length scale, non viable cosmological model can be obtained for the requirement of $Ω_Λ<1/3$ which conflicts with current comic observations. When the future event horizon is taken as the role of cosmological length scale, the forms of effective equation of state of horizon cosmological constants are the the same as the holographic ones. But, their evolutions are different because of the effective interaction with cold dark matter.

astro-ph.CO↗

arXiv2010

Area spectrum of near-extremal SdS black holes via the new interpretation of quasinormal modes

Motivated by the recent work about a new physical interpretation of quasinormal modes by Maggiore, we investigate the quantization of near-extremal Schwarzschild-de Sitter black holes in the four dimensional spacetime. Following Kunstatter's method, we derive the area and entropy spectrum of near-extremal Schwarzschild-de Sitter black holes which differs from Setare's result. Furthermore, we find that the derived a universal area spectrum is $2πn$ which is equally spaced.

astro-ph.CO↗

arXiv2009

Cosmic Constraint on Ricci Dark Energy Model

In this paper, a holographic dark energy model, dubbed Ricci dark energy, is confronted with cosmological observational data from type Ia supernovae (SN Ia), baryon acoustic oscillations (BAO) and cosmic microwave background (CMB). By using maximum likelihood method, it is found out that Ricci dark energy model is a viable candidate of dark energy model with the best fit parameters: $Ω_{m0}=0.34\pm 0.04$, $α=0.38\pm 0.03$ with $1σ$ error. Here, $α$ is a dimensionless parameter related with Ricci dark energy $ρ_{R}$ and Ricci scalar $R$, i.e., $ρ_{R}\propto αR$.

astro-ph↗

arXiv2009

Generalized Holographic and Ricci Dark Energy Models

In this paper, we consider generalized holographic and Ricci dark energy models where the energy densities are given as $ρ_{R}=3c^2M^{2}_{pl}Rf(H^2/R)$ and $ρ_{h}=3c^2M^{2}_{pl}H^2g(R/H^2)$ respectively, here $f(x),g(y)$ are positive defined functions of dimensionless variables $H^2/R$ or $R/H^2$. It is interesting that holographic and Ricci dark energy densities are recovered or recovered interchangeably when the function $f(x)=g(y)\equiv 1$ or$f=g\equiv Id$ is taken respectively (for example $f(x),g(x)=1-ε(1-x)$, $ε=0 \text{or} 1$ respectively). Also, when $f(x)\equiv xg(1/x)$ is taken, the Ricci and holographic dark energy models are equivalents to a generalized one. When the simple forms $f(x)=1-ε(1-x)$ and $g(y)=1-η(1-y)$ are taken as examples, by using current cosmic observational data, generalized dark energy models are researched. As expected, in these cases, the results show that they are equivalent ($ε=1-η=1.312$) and Ricci-like dark energy is more favored relative to the holographic one where the Hubble horizon was taken as an IR cut-off. And, the suggestive combination of holographic and Ricci dark energy components would be $1.312 R-0.312H^2$ which is $2.312H^2+1.312\dot{H}$ in terms of $H^2$ and $\dot{H}$.

astro-ph.CO↗

arXiv2009

Time Variable Cosmological Constants from the Age of Universe

In this paper, time variable cosmological constant, dubbed {\it age cosmological constant}, is investigated motivated by the fact: any cosmological length scale and time scale can introduce a cosmological constant or vacuum energy density into Einstein's theory. The age cosmological constant takes the form $ρ_Λ=3c^2M^2_P/t_Λ^2$, where $t_Λ$ is the age of our universe or conformal time. The effective equation of state of age cosmological constant are $w^{eff}_Λ=-1+{2/3}\frac{\sqrt{Ω_Λ}}{c}$ and $w^{eff}_Λ=-1+{2/3}\frac{\sqrt{Ω_Λ}}{c}(1+z)$ when the age of universe and conformal time are taken as the role of cosmological time scales respectively. They are the same as the so-called agegraphic dark energy models. However, the evolution history are different from the agegraphic ones for their different evolution equations.

astro-ph.CO↗

arXiv2009

Constraints on Kinematic Model from Recent Cosmic Observations: SN Ia, BAO and Observational Hubble Data

In this paper, linear first order expansion of deceleration parameter $q(z)=q_0+q_1(1-a)$ ($M_1$), constant jerk $j=j_0$ ($M_2$) and third order expansion of luminosity distance ($M_3$) are confronted with cosmic observations: SCP 307 SN Ia, BAO and observational Hubble data (OHD). Likelihood is implemented to find the best fit model parameters. All these models give the same prediction of the evolution of the universe which is undergoing accelerated expansion currently and experiences a transition from decelerated expansion to accelerated expansion. But, the transition redshift depends on the concrete parameterized form of the model assumed. $M_1$ and $M_2$ give value of transition redshift about $z_t\sim 0.6$. $M_3$ gives a larger one, say $z_t\sim 1$. The $χ^2/dof$ implies almost the same goodness of the models. But, for its badness of evolution of deceleration parameter at high redshift $z>1$, $M_3$ can not be reliable. $M_1$ and $M_2$ are compatible with $Λ$CDM model at the $2σ$ and $1σ$ confidence levels respectively. $M_3$ is not compatible with $Λ$CDM model at $2σ$ confidence level. From $M_1$ and $M_2$ models, one can conclude that the cosmic data favor a cosmological model having $j_0<-1$.

astro-ph.CO↗

arXiv2009

Cosmic Constraints on Holographic Dark Energy in Brans-Dicke Theory

In this paper, the holographic dark energy in Brans-Dicke theory is confronted by cosmic observations from SN Ia, BAO and CMB shift parameter. The best fit parameters are found in $1σ$ region: $Ω_{h0}=0.683^{+0.035}_{-0.038}$, $c=0.605^{+0.138}_{-0.107}$ and $α=0.00662^{+0.00477}_{-0.00467}$ (equivalently $ω=905.690^{+637.906}_{-651.471}$ which is less the solar system bound and consistent with other constraint results). With these best fit values of the parameters, it is found the universe is undergoing accelerated expansion, and the current value of equation of state of holographic dark energy $w_{h0}=-1.246^{+0.191}_{-0.144}$ which is phantom like in Brans-Dicke theory. The evolution of effective Newton's constant is also explored.

astro-ph.CO↗

arXiv2009

Holographic Dark Energy in Brans-Dicke Theory

In this paper, the holographic dark energy model is considered in Brans-Dicke theory where the holographic dark energy density $ρ_Λ =3c^2 M^{2}_{pl} L^{-2}$ is replaced with $ρ_{h}=3c^2 Φ(t)L^{-2}$. Here $Φ(t)=\frac{1}{8πG}$ is a time variable Newton constant. With this replacement, it is found that no accelerated expansion universe will be achieved when the Hubble horizon is taken as the role of IR cut-off. When the event horizon is adopted as the IR cut-off, an accelerated expansion universe is obtained. In this case, the equation of state of holographic dark energy $w_h$ takes a modified form $w_h=-{1/3}(1+α+\frac{2}{c}\sqrt{Ω_{h}})$. In the limit $α\to 0$, the 'standard' holographic dark energy is recovered. In the holographic dark energy dominated epoch, power-law and de Sitter time-space solutions are obtained.

astro-ph↗

arXiv2009

A CLT for the L^{2} modulus of continuity of Brownian local time

Let $\{L^{x}_{t} ; (x,t)\in R^{1}\times R^{1}_{+}\}$ denote the local time of Brownian motion and \[ α_{t}:=\int_{-\infty}^{\infty} (L^{x}_{t})^{2} dx . \] Let $η=N(0,1)$ be independent of $α_{t}$. For each fixed $t$ \[ {\int_{-\infty}^{\infty} (L^{x+h}_{t}- L^{x}_{t})^{2} dx- 4ht\over h^{3/2}} \stackrel{\mathcal{L}}{\to}({64 \over 3})^{1/2}\sqrt{α_{t}} η, \] as $h\rar 0$. Equivalently \[ {\int_{-\infty}^{\infty} (L^{x+1}_{t}- L^{x}_{t})^{2} dx- 4t\over t^{3/4}} \stackrel{\mathcal{L}}{\to}({64 \over 3} )^{1/2}\sqrt{α_{1}} η, \] as $t\rar\infty$.

math.PR↗