arXiv subjects

Xiaofei Li

Publications and source records attributed to Xiaofei Li.

81 records · Page 5Linked to original sources

arXiv2017

Joint Alignment of Multiple Point Sets with Batch and Incremental Expectation-Maximization

This paper addresses the problem of registering multiple point sets. Solutions to this problem are often approximated by repeatedly solving for pairwise registration, which results in an uneven treatment of the sets forming a pair: a model set and a data set. The main drawback of this strategy is that the model set may contain noise and outliers, which negatively affects the estimation of the registration parameters. In contrast, the proposed formulation treats all the point sets on an equal footing. Indeed, all the points are drawn from a central Gaussian mixture, hence the registration is cast into a clustering problem. We formally derive batch and incremental EM algorithms that robustly estimate both the GMM parameters and the rotations and translations that optimally align the sets. Moreover, the mixture's means play the role of the registered set of points while the variances provide rich information about the contribution of each component to the alignment. We thoroughly test the proposed algorithms on simulated data and on challenging real data collected with range sensors. We compare them with several state-of-the-art algorithms, and we show their potential for surface reconstruction from depth data.

cs.CV↗

arXiv2017

Asymptotic Analysis of the Narrow Escape Problem in Dendritic Spine Shaped Domain: Three Dimension

This paper deals with the three-dimensional narrow escape problem in dendritic spine shaped domain, which is composed of a relatively big head and a thin neck. The narrow escape problem is to compute the mean first passage time of Brownian particles traveling from inside the head to the end of the neck. The original model is to solve a mixed Dirichlet-Neumann boundary value problem for the Poisson equation in the composite domain, and is computationally challenging. In this paper we seek to transfer the original problem to a mixed Robin-Neumann boundary value problem by dropping the thin neck part, and rigorously derive the asymptotic expansion of the mean first passage time with high order terms. This study is a nontrivial generalization of the work in \cite{Li}, where a two-dimensional analogue domain is considered.

math-ph↗

arXiv2017

Electromagnetic interior transmission eigenvalue problem for inhomogeneous media containing obstacles and its applications to near cloaking

This paper is concerned with the invisibility cloaking in electromagnetic wave scattering from a new perspective. We are especially interested in achieving the invisibility cloaking by completely regular and isotropic mediums. Our study is based on an interior transmission eigenvalue problem. We propose a cloaking scheme that takes a three-layer structure including a cloaked region, a lossy layer and a cloaking shell. The target medium in the cloaked region can be arbitrary but regular, whereas the mediums in the lossy layer and the cloaking shell are both regular and isotropic. We establish that there exists an infinite set of incident waves such that the cloaking device is nearly-invisible under the corresponding wave interrogation. The set of waves is generated from the Maxwell-Herglotz approximation of the associated interior transmission eigenfunctions. We provide the mathematical design of the cloaking device and sharply quantify the cloaking performance.

math-ph↗

arXiv2016

Audio-Visual Speaker Diarization Based on Spatiotemporal Bayesian Fusion

Speaker diarization consists of assigning speech signals to people engaged in a dialogue. An audio-visual spatiotemporal diarization model is proposed. The model is well suited for challenging scenarios that consist of several participants engaged in multi-party interaction while they move around and turn their heads towards the other participants rather than facing the cameras and the microphones. Multiple-person visual tracking is combined with multiple speech-source localization in order to tackle the speech-to-person association problem. The latter is solved within a novel audio-visual fusion method on the following grounds: binaural spectral features are first extracted from a microphone pair, then a supervised audio-visual alignment technique maps these features onto an image, and finally a semi-supervised clustering method assigns binaural spectral features to visible persons. The main advantage of this method over previous work is that it processes in a principled way speech signals uttered simultaneously by multiple persons. The diarization itself is cast into a latent-variable temporal graphical model that infers speaker identities and speech turns, based on the output of an audio-visual association process, executed at each time slice, and on the dynamics of the diarization variable itself. The proposed formulation yields an efficient exact inference procedure. A novel dataset, that contains audio-visual training data as well as a number of scenarios involving several participants engaged in formal and informal dialogue, is introduced. The proposed method is thoroughly tested and benchmarked with respect to several state-of-the art diarization algorithms.

cs.CV↗

arXiv2016

Estimation of the Direct-Path Relative Transfer Function for Supervised Sound-Source Localization

This paper addresses the problem of binaural localization of a single speech source in noisy and reverberant environments. For a given binaural microphone setup, the binaural response corresponding to the direct-path propagation of a single source is a function of the source direction. In practice, this response is contaminated by noise and reverberations. The direct-path relative transfer function (DP-RTF) is defined as the ratio between the direct-path acoustic transfer function of the two channels. We propose a method to estimate the DP-RTF from the noisy and reverberant microphone signals in the short-time Fourier transform domain. First, the convolutive transfer function approximation is adopted to accurately represent the impulse response of the sensors in the STFT domain. Second, the DP-RTF is estimated by using the auto- and cross-power spectral densities at each frequency and over multiple frames. In the presence of stationary noise, an inter-frame spectral subtraction algorithm is proposed, which enables to achieve the estimation of noise-free auto- and cross-power spectral densities. Finally, the estimated DP-RTFs are concatenated across frequencies and used as a feature vector for the localization of speech source. Experiments with both simulated and real data show that the proposed localization method performs well, even under severe adverse acoustic conditions, and outperforms state-of-the-art localization methods under most of the acoustic conditions.

cs.SD↗

arXiv2014

A New Model for Solving Narrow Escape Problem in Domain with Long Neck

The narrow escape problem arises in deriving the asymptotic expansion of the solution of an inhomogeneous mixed Dirichlet-Neumann boundary value problem. In this paper, we mainly deal with narrow escape problem in a smooth domain connected to a long neck-Dendritic spine shape domain, which has a certain significance in biology. Since the special geometry of dendritic spine, we develop a new model for solving this narrow escape problem which is Neumann-Robin Boundary Model. This model transform spine singular domain to smooth spine head domain by inserting Robin boundary condition to the connection part between spine head and neck. We rigorously find the high-order asymptotic expansion of Neumann-Robin Boundary Model and apply it to the solution of narrow escape problem in a dendritic spine shape domain. Our results show that the asymptotic expansion of the Neumann-Robin Boundary Model can be easily applied to the narrow escape problem for any smooth spine head domain with straight spine neck. By numerical simulations, we show that there is great agreement between the results of our Neumann-Robin Boundary Model and the original escape problem. In this paper, we also get some results for non-straight long spine neck case by considering curvature of spine neck.

math.AP↗

arXiv2013

Bounds on the size of an inclusion using the translation method for two-dimensional complex conductivity

The size estimation problem in electrical impedance tomography is considered when the conductivity is a complex number and the body is two-dimensional. Upper and lower bounds on the volume fraction of the unknown inclusion embedded in the body are derived in terms of two pairs of voltage and current data measured on the boundary of the body. These bounds are derived using the translation method. We also provide numerical examples to show that these bounds are quite tight and stable under measurement noise.

math.AP↗

arXiv2009

Ion specific effects on phase transitions in protein solutions

A recent Monte Carlo simulation determined the potential of mean force between two lysozyme molecules in various aqueous solutions [M. Lund et al. Phys. Rev. Lett. 100, 258105 (2008)]. The study involved a combination of explicit solvent and continuum model simulations and showed that there are significant ion-specific protein-protein interactions due to hydrophobic patches on the protein surfaces. In this paper we use the results of their study to determine the phase diagram for lysozyme for aqueous solutions of NaCl and NaI. Two of the three phase diagrams have a stable fluid-fluid critical point, while the third has a slightly metastable critical point. This results from a secondary extremum in the potential associated with a repulsive interaction. This repulsive interaction reduces the effective range of the attractive interaction and produces a metastable critical point. We compare the results of one of these phase diagrams with that for a model that includes ion-dispersion forces, but does not contain solvent structural effects.

physics.chem-ph↗

arXiv2005

Simple Model of Sickle Hemogloblin

A microscopic model is proposed for the interactions between sickle hemoglobin molecules based on information from the protein data bank. A Monte Carlo simulation of a simplified two patch model is carried out, with the goal of understanding fiber formation. A gradual transition from monomers to one dimensional chains is observed as one varies the density of molecules at fixed temperature, somewhat similar to the transition from monomers to polymer fibers in sickle hemoglobin molecules in solution. An observed competition between chain formation and crystallization for the model is also discussed. The results of the simulation of the equation of state are shown to be in excellent agreement with a theory for a model of globular proteins, for the case of two interacting sites.

cond-mat.soft↗