基于隶属度的人脸图像特征抽取和识别

本文首先从代数理论和模糊数学导出一种新概念，矢量隶属函数，由此构造图像的隶属函数，并证明它具有一些不变性，提出一种基于隶属函数的图像特征抽取和识别方法，并应用于人脸图像识别问题。实验结果表明，隶属函数具有良好的识别分离能力。     

基于中心矩特征的空间目标识别方法

目标的雷达散射截面（RCS）包含了丰富的目标类别信息，有效地利用目标RCS特征对空间目标的雷达识别具有重要的意义。该文利用空间目标回波的距离维信号来进行识别。中心矩特征具有平移不变性，是一种简单有效的波形特征提取算法。文中首先提取中心矩作为特征向量，再采用Fisher判据进一步进行特征压缩，最后利。用支撑矢量机（SVM）分类算法实现识别。基于实测数据的仿真实验结果表明，该方法具有较好的识别性能和推广能力。     

图像奇异值特征矢量缩放不变性分析及应用

图像奇异值特征矢量以其稳定性和它在转置、平移、旋转和镜像变换下具有的不变性,被认为是一种图像的代数特征,广泛应用于图像匹配和图像识别。通过矩阵运算,证明了奇异值特征矢量在图像做缩放变换时具有不变性,并将这种性质运用于图像匹配。实验结果表明这种不变性与奇异值特征矢量在图像做转置、平移、旋转和镜像变换时所具有的不变性一样,是奇异值特征矢量能够作为图像代数特征的一个重要依据。     

基于小波变换和DCT的字符图像特征抽取新方法

从特征矢量的不变性和抗噪性角度,提出了一种基于小波变换(WT)和离散余弦变换(DCT)的字符特征抽取新方法。利用圆周投影算法,把二维的字符图像转换为一维投影。基于WT和DCT的非线性变换,克服了因变形和噪声引起的一维投影的非线性变形失真。通过对识别不同大小、方向及噪声的字符仿真实验和不同特征抽取方法的对比实验,以及对工业标牌字符的识别,表明该特征抽取方法具有尺度和旋转不变性,有较好的抗噪声能力和很好的分类性能。     

基于颜色矢量角的彩色图像检索算法研究

通过分析利用颜色直方图进行图像检索时存在的问题，提出了一种基于图像边缘空间分布特征的图像检索算法。即利用颜色矢量角对亮度不敏感，对色度和饱和度敏感的特性来提取图像的边缘信息。在图像的边缘空间，提取图像边缘点的颜色矢量角直方图，以此来描述图像的内容。该算法充分利用了图像的颜色信息、边缘信息及形状信息，实验表明该算法具有较好的光照不变性、尺度不变性、平移不变性和旋转不变性。     

基于形心极二阶矩比特征辅助的星图识别算法

在传统的星图识别算法中，本文在三角形算法的基础上，提出一种改进算法，将待匹配三角形的形心极二阶矩比作为导航特征库中的辅助匹配特征量，以增加特征矢量维数提高识别精度，同时采用函数映射搜索方式提高识别速度。     

不变性理论用于空中目标的识别

以计算机视觉的不变性理论为基础，用飞机红外图像边缘中的角点反映飞机的几何结构．然后对飞机的角点应用变化，以构造出两种不变特征，实验表明这两种特征具有一定不变性．推导了Ｍａｈａｌａｎｏｂｉｓ距离的平移、旋转和仿射不变性，并应用到空中目标的识别，实验表明这种方法效果良好，性能稳定     

基于边缘梯度方向直方图的静态手语识别

文中采用边缘梯度方向直方图作为手势的特征矢量进行手语识别,建立归一化的边缘梯度直方图,使用欧氏距离模板匹配法进行手势的特征匹配,手势特征矢量之间的识别速度较快.实验表明:该方法对图像亮度、缩放、平移具有不变性,该方法计算简单、快速,可以用于手语识别系统.     

一种新的雷达目标HRRP识别方法

针对雷达目标回波的非平稳特性,推导了S变换及其快速实现算法,利用电磁场时域有限差分算法仿真了3种军用飞机的宽带散射信号,采用S变换对飞机目标的雷达回波进行时频分析,提取时频分布图的矩特征作为目标特征矢量,利用改进径向基函数神经网络对特征矢量进行训练和学习,最后对3种飞机作了分类识别,取得了很好的识别效果。     

基于拓扑与网格双特征的铭文图形识别方法

青铜器铭文图像有效特征的提取是进行铭文识别的关键步骤,针对以图像为信息载体的铭文特征提取方法由于特征维度高、特征向量复杂而识别准确度低的问题,提出了一种基于拓扑与网格双特征的铭文图形集成学习识别方法.以图形为铭文特征的表征,所提方法提取拓扑特征和7维文字结构图形特征,有效描述了铭文文字的结构信息.在此基础上,所提方法利...     

IC缺陷轮廓分形维估计的小波方法

为了有效的进行集成电路成品率预报和故障分析,与光刻有关的硅片表面缺陷通常被假定为圆形、椭圆或方形.然而,真实缺陷的形貌是多种多样的,它们的形状对集成电路的成品率估计有重要影响.本文讨论了缺陷轮廓所具有的分形特征,并用小波变换的方法对其分形维数进行了估计,估计结果与实例的特征相符,从而为缺陷轮廓的表征和计算机模拟提供了新的特征参数.     

基于高斯曲面特征矩阵的扩展目标形状估计

针对传统方法难以准确估计扩展目标形状的问题,提出一种新的基于高斯曲面拟合的量测模型和基于高斯曲面特征矩阵的形状估计算法。首先,建立能反映目标真实形状的结构点,并产生多个高斯曲面,通过曲面叠加形成任意形状的量测空间分布模型;然后,根据高斯曲面拟合原理,用矩阵表示该拟合曲面主要区域的分布特征,并通过映射方程建立矩阵坐标与笛卡尔坐标的映射关系;最后,通过贝叶斯滤波体系更新拟合矩阵。与现有算法相比,本文算法不需要准确预设目标形状,在量测噪声较大的环境下,可以自适应的拟合目标真实形状。并且,在不需要预设目标形状方程的情况下,可以估计包括空心形状在内的任意不规则目标形状。实验结果表明,在目标初始形状参数不准确的情况下,本文算法正确估计了飞机形状、空心形状和集群目标形状,并且具有较好的扩展目标形状估计性能和较强的鲁棒性。     

基于小波变换的IC缺陷轮廓曲线分形维估计

硅片表面缺陷轮廓的形状特征对集成电路(IC)成品率预报及故障分析有重要影响。讨论了IC缺陷轮廓所具有的分形特征,利用小波变换对其参数曲线的分形维数进行了估计,估计结果与实际特征相符,从而为缺陷轮廓特征的描述和计算机模拟提供了一条新的途径。     

Feature-guided shape-based image interpolation

A feature-guided image interpolation scheme is presented. It is an effective and improved, shape-based interpolation method used for interpolating image slices in medical applications. The proposed method integrates feature line-segments to guide the shape-based method for better shape interpolation. An automatic method for finding these line segments is given. The proposed feature-guided shape-based method can manage translation, rotation and scaling situations when the slices have similar shapes. It can also interpolate intermediate shapes when the successive slices do not have similar shapes. This method is experimentally evaluated using artificial and real two-dimensional and three-dimensional data. The proposed method generated satisfactory interpolated results in these experiments. We demonstrate the practicality, effectiveness and reproducibility of the proposed method for interpolating medical images.     

Real-Time Classification of Multiunit Neural Signals Using Reduced Feature Sets

Classification of characteristic neural spike shapes in multi-unit recordings is performed in real time using a reduced feature set. A model of uncorrelated signal-related noise is used to reduce the feature set by choosing a subset of aperiodic samples which is effective for discrimination between signals by a nearest-mean algorithm. Initial signal classes are determined by an unsupervised clustering algorithm applied to the reduced features of the learning set events. Classification is carried out in real time using a distance measure derived for the reduced feature set. Examples of separation and correlation of multiunit activity from cat and frog visual systems are described.     

Shape control of polymer reflow structures fabricated by nanoimprint lithography

Nanoimprinted resist pre-forms were modified using thermal reflow. This post-processing of binary structures enabled us to generate lens-like 3-D structures with different shapes by time- and surface chemistry controlled spreading. The method was extended to feature dimensions down to 100 nm. Surfactant coated line nanostructures were found to be limited by a maximum aspect ratio for imprinted pre-forms. This powerful post-processing method can be used either directly as post-processing step in production or for the fabrication of 3-D stamp copies with the desired shapes.     

Representing 3D shapes based on implicit surface functions learned from RBF neural networks

We propose to represent the shape of 3D objects using a neural network classifier. The 3D shape is learned from a neural network, where Radial Basis Function (RBF) is applied as the activation function for each perceptron. The implicit functions derived from the neural network is a combination of radial basis functions, which can represent complex shapes. The use of RBF provides a rotation, translation and scaling invariant feature to represent the shape. We conduct experiments on a new prostate dataset and public datasets. Our testing results show that our neural network-based method can accurately represent various shapes.     

Classification of remotely sensed images using decimal coded morphological profiles

In this paper, we propose a novel method for pixel classification of remotely sensed images. The proposed method exploits the spatial information of image pixels using morphological profiles produced by structuring elements of different sizes and shapes. Morphological profiles produced by multiple structuring elements are combined into a single feature by decimal coding. The advantage of proposed feature is that it can effectively utilize the potential of multiple morphological profiles without increasing the complexity of feature space. The proposed approach was tested on remotely sensed images with known ground truths, and performance was improved up to 27 % in the overall accuracy results over existing techniques.     

Biasing Buckling Direction in Shape‐Programmable Hydrogel Sheets with Through‐Thickness Gradients

A photocrosslinkable poly( N , N ′‐diethylacrylamide) copolymer allows for the photolithographic fabrication of hydrogel sheets with nonuniform crosslinking density and swelling ratio. Using this material system, different 3D shapes with nonzero Gaussian curvature K are successfully programmed by prescribing a “metric” defined by in‐plane variations in swelling. However, this methodology does not control the direction of buckling adopted by each positive K feature, and therefore cannot controllably select between different isometric shapes defined by a single metric. Here, by introducing gradients in swelling through the thickness of the gel sheet by tuning the absorption of the UV‐light used for crosslinking, a preferential buckling direction is locally specified for each feature by the direction of UV exposure. By also controlling the strength of coupling between neighboring features, this is shown to be an effective method to program buckling direction of each unit within a canonical corrugated surface shape.