课程文献中心 更多>>

关键词： 格论   数学形态学   脉冲耦合神经网络   矢量排序  

摘要： 图像信息处理是图像分析的关键步骤,是一种低层次的计算机视觉技术。图像信息处理通过分离目标,提取参数和测量参数将原始图像转化成更抽象的形式,便于图像的分析和理解。由于图像信息处理的重要性和困难性,虽然从20世纪70年代起就已经引起人们的高度重视,但是目前还没有找到一个通用的方法和评判图像信息处理结果的客观标准。 神经网络是图像信息处理的重要手段之一。早在80年代,就有人将神经网络应用于图像的恢复上。由于神经网络是基于生物视觉系统,所以具有传统图像处理技术所不具备的优势。而脉冲耦合神经网络又是其中的佼佼者。 本文的主要研究目的在于对脉冲耦合神经网络进行建模,并通过其在图像处理上的应用,对模型进行验证。 首先,简要的回顾一下脉冲耦合神经网络模型和现在图像处理技术的发展,分析脉冲耦合神经网络的脉冲机制和工作原理,指出了在神经元个数较小,连接情况较为简单的情况下,如何预测脉冲耦合神经网络的行为。 其次,将脉冲耦合神经网络引入在彩色图像分割上,利用它的出色特性,选择适当的彩色空间,去除彩色分量之间的相互联系,在彩色空间的各个分量上分别利用脉冲耦合神经网络进行图像分割,然后将分割的结果通过一定的方法融合,这样就得到一种新的彩色图像分割方法。试验得到了很好的效果。这是脉冲耦合神经网络首次应用于多维空间。 再其次,为了研究和建立脉冲耦合神经网络的数学模型,先简要的回顾了数学形态学与脉冲耦合神经网络之间的关系。在此基础之上,使用更加严密的数学工具——格论对脉冲耦合神经网络进行建模,并利用模型分析先前得到的PCNN处理彩色图像分割的方法和试验结果,指出其优点和不足。随后,利用格论中的矢量排序的方法,重新的设计脉冲耦合神经网络的计算方程和神经单元的选择条件,得到更优秀的PCNN彩色图像分割算法。试验的结果跟先前的预计完全符合,从而进一步的验证了模型的正确性。 最后,关于进一步工作的方向进行了简要的讨论。

关键词： 脉冲耦合神经网络   指纹图像   图像分割   图像去噪  

摘要： 由Eckhorn神经元模型得到的脉冲耦合神经网络模型(PCNN)是直接观察猫的视觉皮层神经细胞并模拟其活动而得到的人工神经网络模型。PCNN不同于传统的人工神经网络模型，它是通过模拟视觉皮层神经细胞的活动而建立的神经网络模型，是对真实神经元的简化与近似。PCNN网络模型所具有的链接域特性和动态阈值衰减特性能够使状态相似的神经元同步输出脉冲，这一点充分模拟了哺乳动物视觉皮层神经元的生物特性，因而在图像分割、边缘提取、目标识别等图像处理方而获得了广泛的应用。 本文深入研究脉冲耦合神经网络的基本原理、运动行为及其特性，本论文对原始PCNN模型做了一定程度的简化，在保持PCNN链接域特性和动态阈值衰减特性的基础上减少了神经元模型的一些参数，提出更适用于图像处理的简化型PCNN模型。从理论上说，简化型PCNN可以用于图像处理的各个方面，本文只对灰度指纹图像进行了分割，又对被白噪声污染的二值指纹图像进行了去噪处理，完成了指纹识别系统中两个重要的环节。同时，实验发现简化型PCNN处理其它类型图像同样有效。 实验结果证明，本文提出的方法用于指纹图像的分割和去噪时，性能优于其它算法，满足指纹识别系统的精度要求。简化型PCNN具有很好的亮度和对比度适应性，同一图像的分割效果不受亮度、对比度不同的影响。

关键词： global exponential periodicity   impulsive neural networks   Lyapunov function   Lipschitz activation function  

摘要： By the Lyapunov function method, combined with the inequality techniques, some criteria are established to ensure the existence, uniqueness and global exponential stability of the periodic solution for a class of impulsive neural networks. The results obtained only require the activation functions to be globally Lipschitz continuous without assuming their boundedness, monotonicity or differentiability. The conditions are easy to check in practice and they can be applied to design globally exponentially periodic impulsive neural networks.

关键词： 脉冲   BAM神经网络   周期解   稳定性  

摘要： 首先建立一种含脉冲的双向联想记忆(BAM)神经网络模型,利用分段Lya-punov函数方法,讨论网络的周期振荡性,全局指数稳定性质及其收敛速率,并给出一实例和数值仿真说明所得结论。

关键词： 模糊粗神经网络   图像融合   遗传算法   滤波  

摘要： 为了获得在图像噪声滤除和细节保留两方面更好的兼顾与平衡,提出了一种兼有模糊化神经元和不可微粗神经元的模糊粗神经网络。采用融合了爬山法的遗传学习算法,增强了网络在学习过程中,特别是进化后期阶段对最优解的局部搜索能力。仿真表明,融合了模糊信息和粗糙集信息处理能力的模糊粗神经网络在图像融合滤波方面比BP网络和粗集神经网络具有更好的性能,是一种性能很好的混合智能神经网络。

关键词： UWB脉冲   捕获   BP神经网络   信噪比  

摘要： 根据UWB信号的特性提出一种在低信噪比条件下实现快速捕获的新方法,即利用BP前向神经网络对极窄脉冲信号进行捕获,有效减少了在低信噪比条件下的平均捕获时间。其中着重分析了利用BP神经网络在低信噪比条件下检测窄脉冲信号的可行性以及接收信号的信噪比对检测的影响。并且对仿真出现的问题进行了分析,得出了一些有价值的结论。

关键词： Signal processing  

摘要： The present work evaluates the application of artificial neural networks for pattern recognition of ultrasonic signals using pulse-echo and TOFD (Time of Flight Diffraction) techniques in weld beads. In this study pattern classifiers are implemented by artificial neural network of backpropagation type using MATLAB®. The ultrasonic signals acquired from pulse-echo and TOFD were introduced, separately, in the artificial neural network with and without preprocessing. The preprocessing was only used to smoothen the signal improving the classification. Four conditions of weld bead were evaluated: lack of fusion (LF), lack of penetration (LP), porosity (PO) and non-defect (ND). The defects were intentionally inserted in a weld bead of AISI 1020 steel plates of 20 mm thickness and were confirmed using radiographic tests. The results obtained show that it is possible to classify ultrasonic signals of weld joints by the pulse-echo and TOFD techniques using artificial neural networks. The results showed a performance superior a 72% of success for test. Although the preprocessing of the signal improved the classification performance of the signals acquired by the TOFD technique considerably, the same didn't happen with the signals acquired by the pulseecho technique.

关键词： Adolescent   Child   Clathrin/genetics*   Clathrin/metabolism   Female   Humans   Lymphoma, B-Cell/genetics*   Lymphoma, Large B-Cell, Diffuse/genetics*   Male   Neoplasm Recurrence, Local/genetics*   Protein-Tyrosine Kinases/genetics*   Protein-Tyrosine Kinases/metabolism  

关键词： Cell Count   Cell Count   Cell Division/physiology*   Cell Proliferation   Fibroblasts/pathology*   Hernia/pathology   Hernia/surgery*   Humans   Materials Testing   Microscopy, Electron, Scanning   Polyglactin 910   Polypropylenes*   Prostheses and Implants*   Surface Properties   Surgical Mesh*  

摘要： Background. The biocompatibility of meshes in hernia surgery seems to be influenced markedly by the amount of the selected material and its structure. Fibroblasts play a major key role during the process of mesh incorporation. This study was performed to investigate differences in cell morphology and proliferation of human fibroblasts cultured on different polypropylene meshes. Methods. In the present in vitro study the cellular response of human fibroblasts was investigated by scanning electron microscopy (SEM), comparing three different polypropylene meshes: a newly constructed low-weight and microporous mesh (NK1), a low-weight and macroporous mesh with absorbable polyglactin filaments (Vypro), and a heavy-weight and microporous mesh (BiomeshP1). Human fibroblasts (1,5.105 cells) were incubated with the meshes (each 12 mm(2)) for 6 hours, 5 days, 2,4,6, and 12 weeks. Computer-assisted morphometry of the fibroblast/mesh surface ratio served to reflect the biological cell response. Results. The Vypro mesh showed the significantly highest fibroblast density during the first 6 weeks, but cell growth was nearly exclusively limited to the polyglactin filaments. At 3 months, after reabsorption of the polyglactin, the fibroblast-coated polypropylene mesh surface was only 50% compared to NK1 and BiomeshP1. The morphologic aspect of the fibroblasts on the BiomeshP1 mesh was much more degenerative and unphysiological, compared to NK1 and Vypro, with isolated, single cells instead of a broad, connective growth. The BiomeshP1 showed a significantly higher fibroblast proliferation around the nodes of the mesh compared to the straight filaments. On the NK1 mesh fibroblasts exclusively proliferated on the filaments but not on the pressed mesh surface. Conclusions. The polymer surface and structure appears to be of major importance for the biocompatibility of meshes: human fibroblasts preferably grow on low-weight meshes, thin filaments, and mesh nodes. Heavyweight meshes induce

关键词： 爆炸物检测   脉冲快热中子分析方法(PFTNA)   中子   人工神经网络  

摘要： 研究利用PFTNA方法和人工神经网络来识别爆炸物的方法。该方法利用PFTNA来获取被检测样品的热中子俘获γ谱和快中子非弹性散射γ谱,然后通过已知能谱对人工神经网络进行训练后识别未知样品。利用这种方法,能够快速的对未知样品进行定性分析。