融合边缘与区域信息的水平集分割算法

针对测地线主动轮廓（GAC）模型容易产生边界泄露且对初始位置敏感及局部图像拟合（LIF）模型容易陷入局部极小的问题,提出融合边缘与区域模型的水平集算法。通过设置权值,该算法能自适应地调整GAC模型和LIF模型在融合算法中所占的比例。对不同图像的实验结果表明该算法的迭代收敛速度比GAC模型和LIF模型要快,分割效果明显优于GAC模型和LIF模型。     

基于区域显著性的活动轮廓分割模型

提出一种新的活动轮廓分割模型,结合视觉显著性检测机制自动获取待分割图像中目标物体的先验形状信息,并自适应地构造初始轮廓,从而降低了初始轮廓位置对分割算法的影响.同时实现了活动轮廓模型对图像的自适应分割和自动分割,使得分割结果更符合人类视觉感知特性.实验结果表明,该模型有较好的分割效果,迭代次数少,且运行时间短.     

基于图像层的双水平集图像分割

传统C-V模型可以将待分割图像分割成目标和背景两区域,但无法实现对多目标图像的分割.多相C-V模型能够对多目标图像进行分割,但需要多次迭代,计算量较大.为了解决上述问题,提出一种基于图像层的双水平集分割算法,该算法通过引入背景填充技术来改变图像背景,从而形成新的图像层,双水平集不断地在新的图像层中进行分割,直到所有目标被分割.这样通过双水平集就可以实现对多目标图像的分割.实验结果表明:该算法能够实现多目标分割,且迭代次数较少,同时具有较强的抗干扰能力和较快的收敛速度.     

基于贝塞尔滤波改进的测地活动轮廓图像分割模型

活动轮廓模型广泛应用于图像分割和目标轮廓提取,基于边缘的测地活动轮廓（GAC）模型在提取边缘明显的物体时得到广泛的应用,但GAC演化过程中,迭代次数较多,耗时较长。针对这一问题,结合贝塞尔滤波理论,对GAC模型改进。首先,利用贝塞尔滤波对图像进行平滑处理,降低噪声;其次,基于贝塞尔滤波的边缘检测函数,构建新的边缘停止项,且并入到GAC模型中;最后,在构造的模型中同时加入反应扩散（RD）项以避免水平集重新初始化。实验结果表明,与多个基于边缘的模型相比,所提模型在保证分割结果精确度的同时,提高了时间效率,更适用于实际应用。     

基于区域GAC模型的二值化水平集图像分割算法

针对测地线主动轮廓(GAC)模型进行了改进,提出了一种基于区域的GAC模型.通过构造基于区域统计信息的符号压力函数取代边界停止函数,有效解决了弱边界目标或离散状边界目标的分割问题.该模型采用二值化水平集方法实现,避免了传统实现方法水平集函数需要重新初始化为符号距离函数,从而导致稳定性差、计算量大、实现复杂等缺点.对不同类型图像的试验结果表明:该算法迭代收敛速度比GAC模型传统实现方法明显加快,且可有效防止边界泄漏,分割效果优于传统GAC模型与C-V模型.     

改进的快速FCM 图像分割算法

针对模糊C 均值(FCM)聚类算法在图像分割中存在的对初始类中心敏感且迭代过程中计算量大的问题,提出了一种改进的算法。先通过精简数据集,减少算法迭代的时间;再使用样本密度法得到FCM 分割算法的初始聚类中心,以减少算法收敛所需的迭代次数。实验结果表明,改进后的分割算法较好地解决了类中心的初始化问题,提高了算法的收敛速度和运行速度。     

结合最大方差比准则和PCNN模型的图像分割

脉冲耦合神经网络（PCNN）模型在图像分割方面有着很好的应用。在各项参数确定的情况下,其分割结果的好坏取决于循环迭代次数的多少,而PCNN模型自身无法实现迭代次数的自动判定。为此提出一种结合最大方差比准则的PCNN迭代次数自动判定算法,用于实现图像的自动分割。算法利用最大方差比准则找到图像的最优分割界限,确定PCNN的迭代次数,获得最优图像分割结果,然后利用最大香农熵准则验证分割结果。实验表明：提出的算法实现了PCNN迭代次数的自动判定,提高了PCNN的迭代速度,运行效率优于基于2D-OTSU和基于交叉熵的自动分割算法,图像分割效果良好。     

基于聚类和改进型水平集的图像分割算法

针对医学图像中通常伴有噪声、多目标的问题,传统水平集无法将图像中的多目标完全分割出来,提出了基于抑制式模糊聚类算法的改进型双水平集模型。首先,利用聚类算法对医学图像进行预分割降噪,通过标准化互信息准则(NMI)判断聚类是否达到满意效果,进而改良聚类算法,再由增加惩罚项的改进型双水平集进行二次分割。实验结果表明,该方法能够降低图像的噪声和算法的敏感性,水平集无需重新初始化,大大减少了计算量和迭代次数,该模型能将伴有噪声的多目标医学图像完全分割出来,获得了预期的分割效果。     

一种基于AOS格式的多相水平集快速分割方法

采用迎风格式的水平集算法实现需要在曲线演化过程中重新初始化水平集函数的要求,为保证算法的稳定,时间步长选取较小值,算法运行速度较慢。文中基于无须重新初始化的水平集方法,在算法数值实现中引入AOS半隐格式,对基于不同统计模型的水平集分割算法给出统一的数值实现。以二相水平集分割算法为基础提出一种新的多相水平集分割方法。该方法采用一个水平集函数进行多次演化实现多区域分割,其优点包括：1)采用AOS半隐格式,该格式无条件稳定,可采用较大的时间步长;2)对多个统计模型进行统一处理;3)采用单一的水平集函数进行演化,减少水平集演化方程的数量,算法更加灵活。实验结果表明,该方法具有较快的分割速度,对具有多个区域的图像能够进行较准确的分割。     

凸松弛Chan-Vese模型的快速分割方法

Chan-Vese模型是图像分割模型中效率较高的一种.传统的分割方法解决Chan-Vese模型出现了计算效率低、占用内存大、对于解决结构复杂的模型运行时间长等问题.针对上述问题,提出了FADMM和ACPDM两种新的快速分割方法.基于离散的二值标记函数,将两相分割模型转化为凸优化模型,结合FISTA算法和Chambolle-Pock算法对ADMM和对偶方法进行改进,采用变分的思想,通过引入辅助变量和拉格朗日乘子,交替迭代直至收敛到泛函的极值.实验结果表明,两种方法在保持图像区域边界的条件下,收敛速度可提高两倍以上.     

Optimum extrapolated method in a special case with application to sor method

Given a linear system Ax = b and an iterative method x ^{(m + 1)} = Gx ^(m)+k, m = 0,1,2,…(1) to solve it, we determine analytically the optimum extrapolation factor of the extrapolated method of (1), when all the eigenvalues of G have the same modulus (Section 2). Then using the SOR theory in the case of consistently ordered matrices A and applying the results of Section 2 to the extrapolated SOR (ESOR) method, we show (Section 3), that the globally optimum parameters of it (and also of the AOR method) [2, 11, 14] are recovered.     

Modelling of stationary and growing cracks in FE framework without remeshing: a state-of-the-art review

The recently developed techniques for modelling cracking within the finite element (FE) framework which use meshes independent of the crack configuration and thus avoid remeshing are reviewed. They combine the traditional FE method with the partition of unity method for modelling individual cracks, intersecting or branching cracks, as well as cracks emanating from holes or other internal interfaces. Numerical integration for the enriched elements, linear dependence and the corresponding solution techniques for the discretized system of equations, as well as the accuracy of the crack tip fields are addressed. Future improvements of the techniques as well as their applications are discussed.     

Ru/TiO_2三甲胺传感元件与化学分析法测定鱼鲜度

在Ｒｕ／ＴｉＯ２型三甲胺传感元件研制的基础上,对其测定鱼鲜度与化学分析法测定鱼鲜度进行比较,进一步证明了三甲胺传感元件的实用和开发价值。     

Swarm intelligent approaches to auto-localization of nodes in static UWB networks

In this paper, we address the problem of localizing sensor nodes in a static network, given that the positions of a few of them (denoted as “beacons“) are a priori known. We refer to this problem as “auto-localization.” Three localization techniques are considered: the two-stage maximum-likelihood (TSML) method; the plane intersection (PI) method; and the particle swarm optimization (PSO) algorithm. While the first two techniques come from the communication-theoretic “world,” the last one comes from the soft computing “world.” The performance of the considered localization techniques is investigated, in a comparative way, taking into account (i) the number of beacons and (ii) the distances between beacons and nodes. Since our simulation results show that a PSO-based approach allows obtaining more accurate position estimates, in the second part of the paper we focus on this technique proposing a novel hybrid version of the PSO algorithm with improved performance. In particular, we investigate, for various population sizes, the number of iterations which are needed to achieve a given error tolerance. According to our simulation results, the hybrid PSO algorithm guarantees faster convergence at a reduced computational complexity, making it attractive for dynamic localization. In more general terms, our results show that the application of soft computing techniques to communication-theoretic problems leads to interesting research perspectives.     

RBF网络的微分进化正交最小二乘算法

研究用于径向基函数（RBF）网络训练的一种微分进化正交最小二乘（DEOLS）算法。把微分进化（DE）算法的种群作为正交最小二乘（OLS）算法的候选径向基函数集合,利用OLS对DE的种群个体进行评断,以确定RBF网络的隐结点的数目、中心和宽度。该算法融合了DE的强大搜索能力和OLS的高效评断能力,隐结点的选择比OLS要合理,同时避免DE的复杂性。最后使用实验验证了该算法的优越性。     

Adaptive moving mesh upwind scheme for the two-species chemotaxis model

Chemotaxis systems are used to model the propagation, aggregation and pattern formation of bacteria/cells in response to an external stimulus, usually a chemical one. A common property of all chemotaxis systems is their ability to model a concentration phenomenon—rapid growth of the cell density in small neighborhoods of concentration points/curves. More precisely, the solution may develop singular, spiky structures, or even blow up in finite time. Therefore, the development of accurate and computationally efficient numerical methods for the chemotaxis models is a challenging task.We study the two-species Patlak–Keller–Segel type chemotaxis system, in which the two species do not compete, but have different chemotactic sensitivities, which may lead to a significantly difference in cell density growth rates. This phenomenon was numerically investigated in Kurganov and Luká?ová-Medvi?ová (2014) and Chertock et al. (2018), where second- and higher-order methods on uniform Cartesian grids were developed. However, in order to achieve high resolution of the density spikes developed by the species with a lower chemotactic sensitivity, a very fine mesh had to be utilized and thus the efficiency of the numerical method was affected.In this work, we consider an alternative approach relying on mesh adaptation, which helps to improve the approximation of the singular structures evolved by chemotaxis models. We develop, in particular, an adaptive moving mesh (AMM) finite-volume semi-discrete upwind method for the two-species chemotaxis system. The proposed AMM technique allows one to increase the density of mesh nodes at the blowup regions. This helps to substantially improve the resolution while using a relatively small number of finite-volume cells.     

改进的Newton—Raphson法及其在化工中的应用

本文通过分析指出了一元及多元Newton-Raphson法振荡的一种原因.并提出了一种避免振荡的方法.该方法的主要思想是:(1)不出现振荡时按原始Newton-Raphson法迭代计算;(2)出现振荡时进行搜索以避开振荡区.计算实例表明:本文方法优于原始的Newton-Raphson法.     

A constructive design method for two-layer perceptrons and its applicatio to the design of modular neural networks

Abstract: A multilayer perceptron is known to be capable of approximating any smooth function to any desired accuracy if it has a sufficient number of hidden neurons. But its training, based on the gradient method, is usually a time consuming procedure that may converge toward a local minimum, and furthermore its performance is greatly influenced by the number of hidden neurons and their initial weights. Usually these crucial parameters are determined based on the trial and error procedure, requiring much experience on the designer's part.
In this paper, a constructive design method (CDM) has been proposed for a two-layer perceptron that can approximate a class of smooth functions whose feature vector classes are linearly separable. Based on the analysis of a given data set sampled from the target function, feature vectors that can characterize the function'well'are extracted and used to determine the number of hidden neurons and the initial weights of the network. But when the classes of the feature vectors are not linearly separable, the network may not be trained easily, mainly due to the interference among the hyperplanes generated by hidden neurons. Next, to compensate for this interference, a refined version of the modular neural network (MNN) has been proposed where each network module is created by CDM. After the input space has been partitioned into many local regions, a two-layer perceptron constructed by CDM is assigned to each local region. By doing this, the feature vector classes are more likely to become linearly separable in each local region and as a result, the function may be approximated with greatly improved accuracy by MNN. An example simulation illustrates the improvements in learning speed using a smaller number of neurons.     

十字阵的单基地MIMO雷达中一种低复杂度的二维DOA估计算法

提出了在单基地多输入多输出(Multiple-input multiple-output,MIMO)雷达中的基于十字阵的一种低复杂度的二维波达方向(Direction of arrival,DOA)估计算法。该算法利用传播算子法(Propagator method,PM)避免了协方差矩阵的构造及其特征值分解,也无需谱峰搜索,从而大大降低了运算的复杂度;同时该算法可实现方位角和仰角的自动配对。本文算法的性能在高信噪比下逼近借助旋转不变技术估计信号参数(Estimation ofsignal parameters via rotational invariance techniques,ESPRIT)算法。文中还推导了目标方位角和仰角的均方误差。仿真结果证明了该算法的有效性。     

Inverse analysis method using MPP-based dimension reduction for reliability-based design optimization of nonlinear and multi-dimensional systems

There are two commonly used analytical reliability analysis methods: linear approximation - first-order reliability method (FORM), and quadratic approximation - second-order reliability method (SORM), of the performance function. The reliability analysis using FORM could be acceptable in accuracy for mildly nonlinear performance functions, whereas the reliability analysis using SORM may be necessary for accuracy of nonlinear and multi-dimensional performance functions. Even though the reliability analysis using SORM may be accurate, it is not as much used for probability of failure calculation since SORM requires the second-order sensitivities. Moreover, the SORM-based inverse reliability analysis is rather difficult to develop.This paper proposes an inverse reliability analysis method that can be used to obtain accurate probability of failure calculation without requiring the second-order sensitivities for reliability-based design optimization (RBDO) of nonlinear and multi-dimensional systems. For the inverse reliability analysis, the most probable point (MPP)-based dimension reduction method (DRM) is developed. Since the FORM-based reliability index (β) is inaccurate for the MPP search of the nonlinear performance function, a three-step computational procedure is proposed to improve accuracy of the inverse reliability analysis: probability of failure calculation using constraint shift, reliability index update, and MPP update. Using the three steps, a new DRM-based MPP is obtained, which estimates the probability of failure of the performance function more accurately than FORM and more efficiently than SORM. The DRM-based MPP is then used for the next design iteration of RBDO to obtain an accurate optimum design even for nonlinear and/or multi-dimensional system. Since the DRM-based RBDO requires more function evaluations, the enriched performance measure approach (PMA+) with new tolerances for constraint activeness and reduced rotation matrix is used to reduce the number of function evaluations.