基于迭代交替优化的图像盲超分辨率重建

基于深度卷积神经网络的图像超分辨率重建算法通常假设低分辨率图像的降质是固定且已知的,如双3次下采样等,因此难以处理降质(如模糊核及噪声水平)未知的图像。针对此问题,该文提出联合估计模糊核、噪声水平和高分辨率图像,设计了一种基于迭代交替优化的图像盲超分辨率重建网络。在所提网络中,图像重建器以估计的模糊核和噪声水平作为先验信息,由低分辨率图像重建出高分辨率图像;同时,综合低分辨率图像和估计的高分辨率图像,模糊核及噪声水平估计器分别实现模糊核和噪声水平的估计。进一步地,该文提出对模糊核/噪声水平估计器及图像重建器进行迭代交替的端对端优化,以提高它们的兼容性并使其相互促进。实验结果表明,与IKC, DASR, MANet, DAN等现有算法相比,提出方法在常用公开测试集(Set5, Set14, B100,Urban100)及真实场景图像上都取得了更优的性能,能够更好地对降质未知的图像进行重建;同时,提出方法在参数量或处理效率上也有一定的优势。     

Location update optimization in personal communication systems

Mobility tracking is concerned with finding a mobile subscriber (MS) within the area serviced by the wireless network. The two basic operations for tracking an MS, location updating and paging, constitute additional load on the wireless network. The total cost of updating and paging can be minimized by optimally dividing the service area into location registration (LR) areas. There are various factors affecting this cost, including the mobility and call patterns of the individual MS, the shape, size and orientation of the LR area, and the method of searching for the MS within the LR area. Based on various mobility patterns of users and network architecture, the design of the LR area is formulated as a combinatorial optimization problem. The objective is to minimize the location update cost subject to a constraint on the size of the LR area. This revised version was published online in July 2006 with corrections to the Cover Date.     

Toward energy‐proportional Internet core networks: an energy‐minimized routing and virtual topology design for Internet protocol layer

With the exponential growth of Internet traffic, the energy consumption issue of core networks is increasingly becoming critical. Today's core networks are highly underutilized most of the time because of the over‐provisioning and redundancy dimensioning, which results in severe energy inefficiency. In previous work, many non‐deterministic polynomial‐time hard mathematics formulation models have been proposed to minimize the energy consumption of core networks. However, effective heuristics are needed to solve these models in medium/large‐size networks. This work studies the energy‐minimized routing and virtual topology design problem of the power‐hungry Internet protocol (IP) layer in core networks, aiming to achieve an energy‐proportional IP layer by exploiting the variation of traffic with hours to reconfigure virtual topology and reroute traffic. We formulate energy‐minimized routing and virtual topology design as an Integer linear programming problem and propose a LR algorithm, a heuristic based on the Lagrangian relaxation, to solve this problem in a polynomial‐time. The simulation results indicate that the LR algorithm outperforms the best previous algorithm and can achieve a near energy‐proportional IP layer with significant power saving. Furthermore, a detailed analysis of simulation results is conducted, which suggests a design principle of network equipment to facilitate the power saving. Copyright © 2013 John Wiley & Sons, Ltd.     

On the performance of cyclic moments-based parameter estimators ofamplitude modulated polynomial phase signals

Parameter estimation of amplitude-modulated polynomial phase signals embedded in additive white Gaussian noise is considered. The amplitude modulation is modeled as the sum of a real-valued deterministic function and a zero mean correlated stationary random process. It is shown that cyclic moments-based estimators, previously proposed for parameter estimation of polynomial phase signals modulated by stationary random processes, can be adapted to the more general signal model considered here. The covariance matrices of the cyclic moments-based amplitude and phase parameter estimators are derived for large sample lengths. Using this result, it is shown how the lags can be chosen to minimize the large-sample variances of the cyclic moments-based phase parameter estimators. Comparisons with the Cramer-Rao bounds are performed under the assumption of a Gaussian modulating process. The theoretical derivations are confirmed by simulation results     

信号估计中的核回归方法

核回归方法是比较常用的一种非参数估计方法。讨论了核回归方法在一维信号估计中的理论与应用,实验比较了高斯核函数的平滑参数h及多项式阶数N对估计效果的影响。结果是在相同阶数N下,较小的h使所有的估计点都收敛到观察值,反之则是一个N阶多项式拟合。在相同h下,阶数N越大,误差越小,计算量也较大,但重构效果的提升并不明显。     

Design and performance analysis of wavelength/time (W/T) matrix codes for optical CDMA

Two-dimensional (2-D) codes for optical CDMA (OCDMA) are increasingly important because the code set size (cardinality) of such codes is large and the codes have good spectral efficiency, especially when compared to linear or direct sequence codes. As an example, the 2-D codes described in this paper (that use intensity modulation and direct detection, IM/DD) have a cardinality of 32, and their spectral efficiency is /spl sim/0.5 bit/s/Hz when a guard-time is used to avoid intersymbol interference. The cardinality is readily increased to 64-80, using the techniques described in the paper. The next best 2-D codes of comparable cardinality that use IM/DD tend to have a lower spectral efficiency (going like 1/K, where K is the cardinality) because they do not support multiple entries per row or per column of the code matrix. To improve on the spectral efficiency of the codes described in this paper, bipolar codes must be considered. Two-dimensional codes or matrices can be generated from pseudoorthogonal (PSO) sequences by means of simple quasigraphical operations. Important results of this construction are that both the cardinality and the spectral efficiency or information spectral density of the set of matrices is higher than that of the generating set of sequences. The matrices can be interpreted (implemented) as space/time (S/T) or wavelength/time (W/T) matrix codes for OCDMA applications. The resultant matrix codes are robust, have high information spectral density, and are effective wavelength multipliers. This paper describes the design and construction of the matrices; analyzes their performance from a communications viewpoint; describes their use as codes for the asynchronous, concurrent communication of multiple users; and analyzes the bit error rate performance based on capturing and modeling a typical network topology and performing a numerical modeling of the system.     

Estimation of continuous-time autoregressive model from finelysampled data

We extend our two earlier continuous-time estimation methods for continuous-time autoregressive (CAR) model to derive estimators using only finely sampled discrete-time data. The approach is based on the approximation of derivatives by divided differences, coupled with some bias correction. Two types of estimators are provided, having bias of the order O(h) or of O(h²) respectively, for small sampling interval h. The procedures are computationally efficient and always yield a stable autoregressive polynomial. Simulations show that their bias are quite low     

Frequency-selective fading channel estimation with a polynomialtime-varying channel model

A rectangular-windowed least-squares estimator using a polynomial model of the time-varying channel taps is proposed for estimating the impulse response of a frequency-selective fading channel. This method provides a significant improvement in mean square identification error (MSIE) over the conventional least mean squares (LMS) and the exponentially weighted recursive least squares (EW-RLS) algorithms without a polynomial model. A detailed study of the effects of channel parameters, such as the fading rate and the signal-to-noise ratio, on the proposed method is carried out. The performance of the method depends on the window size of the least squares estimator and the polynomial order being used. Algorithms to obtain the approximately optimal window size and polynomial order are proposed and are shown to perform well     

A spatially adaptive nonparametric regression image deblurring.

We propose a novel nonparametric regression metthod for deblurring noisy images. The method is based on the local polynomial approximation (LPA) of the image and the paradigm of intersecting confidence intervals (ICI) that is applied to define the adaptive varying scales (window sizes) of the LPA estimators. The LPA-ICI algorithm is nonlinear and spatially adaptive with respect to smoothness and irregularities of the image corrupted by additive noise. Multiresolution wavelet algorithms produce estimates which are combined from different scale projections. In contrast to them, the proposed ICI algorithm gives a varying scale adaptive estimate defining a single best scale for each pixel. In the new algorithm, the actual filtering is performed in signal domain while frequency domain Fourier transform operations are applied only for calculation of convolutions. The regularized inverse and Wiener inverse filters serve as deblurring operators used jointly with the LPA-design directional kernel filters. Experiments demonstrate the state-of-art performance of the new estimators which visually and quantitatively outperform some of the best existing methods.     

针对RDF概率图查询的基数估计方法

资源描述框架图查询中,准确估计查询结果的大小是查询优化器中的关键步骤.已有方法忽略了该图自身的不确定性以及子查询间的关联关系,无法有效估计结果.针对该问题,本文提出一种基于贝叶斯模型的基数估计方法.该方法引入贝叶斯网络模型,挖掘出子查询内的属性依赖.同时,在这些属性依赖的基础上提出子网拼接方法,计算出子查询间的影响因子.最后,利用以上信息准确估计出任意查询结果集的基数.实验表明:与已有方法相比,本文方法的准确性提高15%以上,性能没有大幅度下降.     

High resolution image formation from low resolution frames using Delaunay triangulation

An algorithm based on spatial tessellation and approximation of each triangle patch in the Delaunay (1934) triangulation (with smoothness constraints) by a bivariate polynomial is advanced to construct a high resolution (HR) high quality image from a set of low resolution (LR) frames. The high resolution algorithm is accompanied by a site-insertion algorithm for update of the initial HR image with the availability of more LR frames till the desired image quality is attained. This algorithm, followed by post filtering, is suitable for real-time image sequence processing because of the fast expected (average) time construction of Delaunay triangulation and the local update feature.     

Online Stator and Rotor Resistance Estimation Scheme Using Artificial Neural Networks for Vector Controlled Speed Sensorless Induction Motor Drive

This paper presents a new method of online estimation for the stator and rotor resistances of the induction motor for speed sensorless indirect vector controlled drives, using artificial neural networks. The error between the rotor flux linkages based on a neural network model and a voltage model is back propagated to adjust the weights of the neural network model for the rotor resistance estimation. For the stator resistance estimation, the error between the measured stator current and the estimated stator current using neural network is back propagated to adjust the weights of the neural network. The rotor speed is synthesized from the induction motor state equations. The performance of the stator and rotor resistance estimators and torque and flux responses of the drive, together with these estimators, are investigated with the help of simulations for variations in the stator and rotor resistances from their nominal values. Both resistances are estimated experimentally, using the proposed neural network in a vector controlled induction motor drive. Data on tracking performances of these estimators are presented. With this speed sensorless approach, the rotor resistance estimation was made insensitive to the stator resistance variations both in simulation and experiment. The accuracy of the estimated speed achieved experimentally, without the speed sensor clearly demonstrates the reliable and high-performance operation of the drive     

一种微波多工器的网络综合方法研究

在滤波器多项式综合基础上,将多项式综合技术在多工器上做了推广,将传统的多项式综合技术从传统的双端口网络拓展到多端口网络,并且给出综合实例。该方法是在归一化频率中综合考虑各路滤波器之间的相互影响,运用迭代算法综合得到多工器的的特征多项式。最终综合结果满足设计指标,同时也验证了这种方法的可行性。     

Classifier shared deep network with multi-hierarchy loss for low resolution face recognition

Face images in real Closed-Circuit Television (CCTV) are usually with low resolution, which remarkably deteriorates the performance of existing face recognition algorithms and hinders the application of face recognition. The main technical focus of this issue, matching between high-resolution (HR) and low-resolution (LR) face images has attracted significant attention. In order to better address this problem, we propose a Classifier Shared Deep Network with Multi-Hierarchy Loss (CS-MHL-Net) for low-resolution face recognition (LRFR) in this paper. Firstly, considering that contrastive loss and its variants are not conducive to the convergence of network and the reduction of discrepancy, a shared classifier between HR and LR face images is proposed to further narrow the domain gap between HR and LR by sharing the corresponding weights which can be seen as the class center. Secondly, to fully exploit intermediate features and loss constraints, we embed multi-hierarchy loss into intermediate layers, with the target of reducing the distances between HR and LR intermediate features after max pooling and avoiding the decreasing of accuracy caused by over-utilization of intermediate features. Experimental results on LFW and SCface demonstrate the effectiveness and superiority of the proposed method.     

Asymptotically optimum estimation of signal waveform in the presence of uncertainties about the steering vector

The problem of signal waveform estimation using an antenna array in case of uncertainties about the steering vector is considered. New asymptotically (in sample size) optimum estimators are derived. In contrast to the known optimal solutions, the employed method of synthesis yields non-iterative direct-form estimators. Simulation results are provided to evaluate the performance of the synthesized estimators. It is shown that the proposed asymptotic estimators perform as well as the iterative optimal estimators and they outperform the MVDR estimator in a wide range of input signal-to-noise and uncertainty ratios.     

Paging strategy optimization in personal communication systems

Mobility tracking is concerned with finding a mobile subscriber (MS) within the area serviced by the wireless network. The two basic operations for tracking an MS, location updating and paging, constitute additional load on the wireless network. The total cost of updating and paging can be minimized by optimally dividing the cellular area into location registration (LR) areas. In current systems broadcast paging messages are sent within the LR area to alert the MS of an incoming call. In this paper we propose a selective paging strategy which uses the MS mobility and call patterns to minimize the cost of locating an MS within an LR area subject to a constraint on the delay in locating the MS. This revised version was published online in July 2006 with corrections to the Cover Date.     

Long-correlation image models for textures with circular and elliptical correlation structures

A class of random field model having long-correlation characteristics is introduced. Unlike earlier approaches in long-correlation models, the correlation structure is isotropic or elliptical in this new class of random field model. The new model has an advantage of modeling diverse real textures with less than five model parameters. Further, the model parameters match with intuitive attributes of textures, such as smoothness, pattern size, elongation or orientation of patterns. The new long-correlation models are based on the fractional differencing of a two-dimensional (2-D) autoregressive polynomial defined by eight symmetric neighbors, and they are either persistent or periodic models depending whether the roots of the polynomial are real or imaginary. A comprehensive three-step algorithm for parameter estimation is developed, and the statistical properties of the estimators are also discussed. The validity of the new model in modeling textures is tested by synthesizing images from manually selected parameters as well as parameters estimated from real textures. It is shown that an image with desired attributes can be synthesized by selecting proper values of the parameters. Further, it is shown that the models introduced can be used in modeling wide range of textures by synthesizing images resembling real textures from estimated parameters.     

Enumeration of all simple paths in a communication network

An algorithm has been developed for enumerating all simple paths in a communication network for specified terminal pairs. The efficiency of the method increases with the number of terminal pairs for which path enumeration is required. The algorithm generates the paths in ascending order of cardinality. The method is applicable to the graphs having finite non zero failure probability nodes. A FORTRAN program for implementation of the algorithm has been developed and results are reported.     

基于循环生成对抗网络的超分辨率重建算法研究

为了提高图像超分辨率重建的效果,该文将注意力机制引入多级残差网络(Multi-level Residual Attention Network,MRAN)作为CycleGAN的重建网络,提出了基于循环生成对抗网络(CycleGAN)的超分辨率重建模型MRA-GAN.MRA-GAN模型中重建网络负责将低分辨率(LR)图像...