基于模糊距离变换的颗粒图像分割方法

在颗粒图像中,经常存在颗粒重叠或者粘连现象。本文提出了一种新的图像分割方法。该方法首先对颗粒图像进行二值化,然后对图像作模糊距离变换得到模糊距离图,再利用分水岭算法将粘连或者部分重叠颗粒分离。实验证明,此法能够将粘连颗粒物图像有效分离。     

基于自适应多重欧氏距离变换的分水岭粘连颗粒分割方法

颗粒形状以及颗粒粘连程度是影响粘连颗粒分割效果的主要因素,现有粘连颗粒分割方法主要存在两方面不足:一些方法只能应对某一特定形状颗粒的分割问题;大多数方法不能根据颗粒粘连程度自适应地调整分割算法,当粘连程度较大时容易出现欠分割.针对上述问题,该文提出了一种不限定颗粒形状的自适应粘连颗粒图像分割方法,该方法根据颗粒粘连程度...     

粘连颗粒分割方法研究

颗粒图像的分割中,经常遇到颗粒粘连、重叠在一起的现象,需要将它们分离为单个颗粒目标。本文提出了一种基于灰度形态学重建的图像分割新算法,实验结果表明,该算法既能更有效地分割粘连颗粒,其应用于图像处理也有着较好的应用前景。     

一种粘连颗粒图像的快速分割方法

对于工业和医学中常见的颗粒图像，提出了一种快速分割粘连颗粒的方法。利用颗粒图像的轮廓特征，找到若干拐点，从拐点中进一步提取出分割点。然后利用分割点的相互距离等因素，为每个分割点找到相匹配的分割点。连接每对分割点即可分开相粘连的颗粒。这种方法快速有效，取得了良好的效果。     

重叠颗粒图象的分离

在细胞图象、纳米材料图象等颗粒图象中，经常会有颗粒重叠粘连的现象，这给图象后续的分析处理带来了许多困难，针对这一情况，提出一种有效的分离算法，应用于二值化后的数字图象，先对图象作距离变换，使颗粒群中的每个颗粒在距离图象中都产生一个局部最大值，再利用分水岭算法，从一个初始的高阈值逐渐降低，就能找到边界点，从而将颗粒样分开。     

基于H-顶面积重构的颗粒分割研究与实现

在实现粘连颗粒的分离时,针对分水岭算法中难以确定对物体进行准确标识等问题,综合利用距离变换以及H-顶面积重构先对物体进行准确标识,再使用分水岭算法找到粘结边界点,从而将颗粒分离。本算法在实际应用中取得了很好的效果。     

基于模糊距离变换的岩心图像颗粒分割算法

岩心CT图像中相邻颗粒之间存在相互粘连,粒度参数分析等需要对粘连颗粒进行分割。结合岩心粘连颗粒的特性,提出一种基于模糊距离变换的改进颗粒分割算法。首先,对预处理后的岩心图像进行模糊距离变换并提取出距离信息的灰度图像,然后利用形态学膨胀重构方法提取标记作为后续分水岭算法的种子点,根据种子点采用一种基于测地重建的改进分水岭算法得到相邻种子点之间的分割线,最后完成粘连颗粒的分割。实验结果表明本文算法可以提高粘连颗粒分割的准确度并减轻分水岭算法的过分割现象。     

多细胞粘连的形状识别方法

多细胞粘连识别是图像识别领域的瓶颈问题,为此,提出一种多细胞粘连的形状识别方法。通过改进的阈值分割和八方向边界描述法对细胞进行定位,基于多边形夹角法分离各个粘连细胞,解决了多细胞粘连的形状识别问题。实验结果表明,该方法适用于多细胞复杂粘连情况下的细胞分离和各个细胞的形状识别,系统粘连分离精度为95.3%,形状识别精度达到97.6%,每个细胞识别时间为0.03 s。     

基于分水岭和区域面积加权的粘连枸杞分级方法

针对枸杞分级过程中出现的粘连枸杞分级效率不高、准确性低的缺点,提出了一种基于形态学分水岭和区域面积加权的粘连枸杞分级方法.对粘连的枸杞图像进行形态学预处理消除枸杞烘干晾晒过程中产生的细小噪声,在保持区域轮廓位置不变的同时,尽可能地消除不规则边缘;运用标记极小值的分水岭算法分割图像,依据枸杞红色分量的分布剔除霉变颗粒,对正常枸杞二值图像运用区域面积标记算法扫描标定各个目标获取颗粒的面积,并用霍特林算法获取颗粒的长宽比;以长宽比作为面积的权值对面积加权修正后进行聚类分析,分成3类.实验结果表明:该方法能够较快速、准确地对不同大小的粘连枸杞颗粒进行分类.     

基于形态学重建颗粒分析的微囊体测量方法

电子墨水的制作过程中,微囊体颗粒的大小及均匀性是衡量其质量的重要指标,尽管可通过形态学颗粒分析等手段统计出这些参数,但由于颗粒分析是基于形态学开运算的,在测量过程中,容易造成微囊体颗粒的形状和大小改变。为此提出了一种基于形态学重建颗粒分析的微囊体图像测量方法,首先将获取的彩色微囊体颗粒图像转化为灰度图像,然后利用形态学区域开滤除小噪声斑点后,采用双阈值进行二值化;通过分水岭变换分离相互粘连的微囊颗粒体,最后采用基于形态学开重建的颗粒分析,在有效保证微囊体颗粒形状和大小的前提下,准确统计出微囊体颗粒的大小分布情况。实验表明,该方法能准确有效地统计出微囊体颗粒的大小分布等指标。     

A Newton-Schur alternative to the consistent tangent approach in computational plasticity

We introduce a Newton-Schur alternative to the consistent tangent approach in plasticity computations and show that the structure of the consistent tangent approach is preserved in the new Newton-Schur approach. However, the local constitutive iterations used in the conventional computational plasticity analyses are eliminated when using the Newton-Schur approach. Two numerical examples are presented which illustrate the efficiencies gained by the new approach.     

基于J(o)sang信任模型的信任传递与聚合研究

研究当前信任管理中的信任传递与信任聚合问题,提出了信任传递与信任聚合应遵循的规则.以J(o)sang信任模型为例,分析其信任传递与聚合方法的不足,进而提出了新的信任传递方法和信任聚合方法.分别通过实验和例子与J(o)sang的信任传递和聚合方法进行了对比,结果表明,新信任传递方法更具一般性,而新信任聚合方法更满足信任聚合的规则.

Abstract：

Trust transitivity and aggregation of trust management are studied, and trust transitivity and aggregation rules are presented. Shortcomings of trust transitivity and trust aggregation methods in J(o)sang' s trust model are presented. A new trust transitivity approach is proposed for trust transitivity, and an approach is provided for trust aggregation. By the experiments and examples of the comparision of J(o)sang's approach and aggregation method, the new trust transitivity approach is more general, and the new trust aggregation approach satisfies the rules of trust aggregation.     

The artificial intelligence approach to pattern recognition—a perspective and an overview

The present paper reviews the techniques for automated extraction of information from signals. The techniques may be classified broadly into two categories—the conventional pattern recognition approach and the artificial intelligence (AI) based approach. The conventional approach comprises two methodologies—statistical and structural. The paper reviews salient issues in the application of conventional techniques for extraction of information. The systems that use the artificial intelligence approach are characterized with respect to three key properties. The basic differences between the approaches and the computational aspects are reviewed. Current trends in the use of the AI approach are indicated. Some key ideas in current literature are reviewed.     

汉语文语转换系统中停顿指数的自动标注

本文采用了一个基于C-TOBI的停顿指数标注的语料库,利用有指导的学习方法对自动停顿指数标注方面做了一些有益的探索。本文共实现了三种方法:基本的马尔科夫模型,引入了词长信息的马尔科夫模型,引入词长信息的马尔科夫模型结合基于转换的错误驱动的学习方法。然后通过对3000句的真实文本进行开放测试,以基本的马尔科夫模型的结果作为基准,实验结果不断改进,最终达到了78.6%的准确率,错误代价降低了14.5%。     

Stability analysis of switched systems under dynamical dwell time control approach

This article investigates the stability of a class of switched systems using dynamical dwell time approach. First, the condition for stability of switched systems whose subsystems are stable are presented with dynamical dwell time approach, which is shown to be less conservative in switching law design than dwell time approach. Then the proposed approach is extended to the switched systems with both stable and unstable subsystems. Finally, some numerical examples are given to illustrate the effectiveness of the proposed results.     

A genetic algorithm approach for the cutting stock problem

In this paper, a genetic algorithm approach is developed for solving the rectangular cutting stock problem. The performance measure is the minimization of the waste. Simulation results obtained from the genetic algorithm-based approach are compared with one heuristic based on partial enumeration of all feasible patterns, and another heuristic based on a genetic neuro-nesting approach. Some test problems taken from the literature were used for the experimentation. Finally, the genetic algorithm approach was applied to test problems generated randomly. The simulation results of the proposed approach in terms of solution quality are encouraging when compared to the partial enumeration-based heuristic and the genetic neuro-nesting approach.     

Computing constant offsets of a NURBS B-Rep

An approach for the automatic offset of a NURBS B-Rep has been presented which can be used for a class of manifold B-Reps. The approach offsets each of the trimmed surfaces (faces) of the B-Rep and then removes the gaps and intersections between offset faces automatically, if any. The offset B-Rep is then created by sewing all the updated offset faces. The practical approach and the treatment of the subject presented are unique and have not been reported earlier. The approach works under the assumption that the number of faces in both base and offset B-Reps is the same. It further assumes that the faces are at least G₁ continuous and non-self intersecting after constant offset. The present approach has given accurate and robust offsets of many lay-up surfaces (B-Reps) in the composite laminate modeling. The approach has better error control. Some of the offsets of lay-up surfaces that are generated using this approach have been presented.     

Frequency-domain approach to the regulation of linear stochastic systems

This paper presents a unification of methodology in the frequency-domain approach to the regulation of linear stochastic time-invariant systems, in which the stochastic processes involved are not necessarily second-order stationary. Various well-known techniques which are currently of interest to statisticians are shown to be special cases. By employing the powerful technique of evolutionary spectral analysis to update the “old” frequency-domain approach, this paper demonstrates that a workable frequency-domain approach can prove to be a reasonable alternative to the time-domain approach.     

基于贝叶斯理论框架的传感器选择算法

针对大规模传感器网络(Large-scale sensor networks)的目标跟踪问题, 本文在贝叶斯(Bayes)框架下, 提出了一种全新的基于传感器选择的多传感器目标跟踪算法.算法的具体思路为:首先基于Bayes框架, 根据不同的管理目标, 推导出传感器选择的目标函数; 然后根据目标函数, 计算出相应的传感器选择方案; 最后将选择的传感器进行数据融合, 求得传感器网络的目标跟踪结果.相比传统的基于量测野值点剔除思想的目标跟踪算法以及基于系统偏差估计的传感器配准算法, 本文提出的基于传感器选择的多传感器目标跟踪算法不仅目标跟踪精度更高, 且跟踪性能更稳定.同时本文提出的传感器选择算法还可以适用于杂波数目较少的目标跟踪场景.仿真结果说明了本文所提算法的有效性.     

面向编程的方法在计算机图形学教学中的应用

该文针对计算机图形学教学中存在的实际问题,提出了面向编程的教学方法.介绍了该方法的主要内容,并对其教学效果进行分析,该方法有效地提高了学生的学习兴趣,使教学质量明显提高.