一种加权矩形提取表情特征的方法*

表情特征提取是面部表情识别的一个关键步骤。针对目前特征提取效率低的情况,通过分析Gabor特征提取的性质和积分图像计算效率高的特点,提出一种利用积分图像技术和矩形模板计算面部特征点特征的方法,模板模拟Gabor的多尺度性,每个模板定义相应的权值,表情图像按照Gabor的各个方向旋转,使用旋转图像积分图和加权模板而不是在积分图上旋转模板提取特征点的特征值,最后将此特征值用于表情分类。实验结果表明,该方法在识别结果相当的情况下极大地提高了特征提取的效率。     

结合空间显著性的融合单演二元模式面部情感分类

针对图像Gabor变换计算代价和存储空间开销较高的问题, 提出一种基于单演信号分析的人脸表情描述方法。该方法首先采用单演信号分析将人脸图像分解为单演幅度、相位和方向三个图像, 并将其划分为多个矩形块子区域; 然后在三幅图像的子区域上提取相应的由空间显著性加权的单演幅度、相位和方向二元模式特征直方图; 最后将结合了空间显著性的三个加权特征进行融合增强特征的可分辨性。在JAFFE人脸表情数据库上的实验结果表明, 该方法能有效提取人脸表情特征, 提高人脸表情的识别率。与基于Gabor的特征相比, 提出的方法具有更高的准确率和较低的特征维度。     

非合作面部晃动情况下的心率检测

目的 心率是直接反映人体健康的重要指标之一,基于视频的非接触式心率检测在医疗健康领域具有广泛的应用前景。然而,现有的基于视频的方法不适用于复杂的现实场景,主要原因是没有考虑视频中目标晃动干扰和空间尺度特征,使得血液容积脉冲信号提取不准确,检测精度不尽人意。为了克服以上缺陷,提出一种抗人脸晃动干扰的非接触式心率检测方法。方法 本文方法主要包含3个步骤：首先,针对目标晃动干扰人脸区域选择的问题,利用判别响应图拟合检测参考图像的人脸区域及主要器官特征点,在人脸跟踪时首次引入倾斜校正思想,输出晃动干扰抑制后的人脸视频;然后,结合空间尺度的差异,采用颜色放大方法对晃动干扰抑制后的人脸视频进行时空处理,提取干净的血液容积脉冲信号;最后,考虑到小样本问题,通过傅里叶系数迭代插值的频域分析方法估计心率。结果 在人脸静止的合作情况以及人脸晃动的非合作情况下采集视频,对心率检测结果进行定量分析,本文方法在两种情况下的准确率分别为97.84%和97.30%,与经典和最新的方法相比,合作情况准确率提升大于1%,非合作情况准确率提升大于7%,表现了出色的性能。结论 提出了一种基于人脸视频处理的心率检测方法,通过有效分析人脸的晃动干扰和尺度特性,提取到干净的血液容积脉冲信号,提高了心率检测的精度和鲁棒性。     

疲劳驾驶面部表情识别算法

针对疲劳驾驶的六种表情 ,提出几何规范化结合 Gabor滤波提取表情特征 ,使用支持向量机对疲劳驾驶的面部表情分类识别的系统。首先对视频图像预处理进行几何规范化 ,利用二维 Gabor核函数构造最优滤波器 48个,获取 48个面部表情特征点 ,最后利用支持向量机进行面部表情分类识别。实验结果表明径向基函数的 SVM性能最好。     

机器人系统中人脸特征提取技术的研究与实现

该文描述了在智能机器人系统中人脸特征提取技术的研究与实现,提出了一种新的并且在机器人系统中实现的人脸特征提取方法,该方法首先利用基于Adaboost的人脸检测算法对采集到的原始图像进行人脸检测,从而得到人脸图像;然后让人脸图像通过一个空间掩模滤波器,去除图像中明显非人脸特征的区域,再经过二值化后得到二值化图像;将二值化图像与一个矩形模板相卷积,得到卷积值与模板索引数的二维曲线图,在二维曲线图中,最高的两个峰就分别对应了眼睛和眉毛,再根据人脸特征几何分布关系判断出眼睛,眉毛和嘴,从而得到最终的人脸特征.该方法检测率高,计算量小,实时性很强,满足了机器人系统中资源有限的约束条件.     

一种结合肤色及类人脸特征的人脸检测

人脸特征提取是人脸检测的关键环节,有效的人脸特征将使得人脸检测更精确。Haar-Like特征作为一种矩形特征,虽然简单、计算迅速,但只能描述特定方向的图形结构。提出的类人脸特征是一种反映人脸灰度分布模型的矩形特征,更加有效地描述了人脸的特征。所提出的人脸检测算法,应用BP神经网络算法训练肤色区域,进行肤色分割。应用类人脸特征的AdaBoost算法进行人脸检测。实验结果表明,该算法可以提高人脸检测的检测率。     

基于非线性Boosting回归的多视角人脸配准

人脸配准可以作为表情分析、人脸识别等人脸相关研究的预处理步骤, 是人脸相关的计算机视觉研究的关键问题. 本文针对图像中水平视角在正负45°内的人脸配准问题, 利用基于Haar特征的非线性Boosting回归算法, 根据标定点邻域内的局部纹理预测标定点的位移, 提出了一种新的基于经典活动形状模型(Active shape model, ASM)的实时多视角人脸配准算法. 在两个数据集合上的测试实验表明, 该算法在速度、准确度和稳定性上都比经典的ASM算法有显著提高且优于近期的改进算法, 具有明显的实用价值.     

Constant Approximation Algorithms for Rectangle Stabbing and Related Problems

In this paper we present constant approximation algorithms for two NP-hard rectangle stabbing problems, called the weighted rectangle stabbing (WRS) problem and the rectangle stabbing with rejecting cost (RSRC) problem. In the WRS problem a set of axis-aligned rectangles is given, with each rectangle associated with a positive weight, and a set of weighted horizontal and/or vertical stabbing lines is sought so that each rectangle is intersected by at least one stabbing line with a weight (called cost) no less than that of the rectangle and the total cost (or weight) of all stabbing lines is minimized. In the RSRC problem each rectangle is associated with an additional positive rejecting cost and is required to be either stabbed by a stabbing line or rejected by paying its rejecting cost. For the WRS problem, we present a polynomial time 2e-approximation algorithm, where e is the natural logarithmic base. Our algorithm is based on a number of interesting techniques such as rounding, randomization, and lower bounding. For the RSRC problem, we give a 3e-approximation algorithm by using a simple but powerful LP rounding technique to identify those to-be-rejected rectangles. Our techniques are quite general and can be easily applied to several related problems, such as the stochastic rectangle stabbing problem and polygon stabbing problem from fixed directions. Algorithms obtained by our techniques are relatively simple and can be easily implemented for practical purpose.     

Finding best-fitted rectangle for regions using a bisection method

In this paper, we have presented a new method for computing the best-fitted rectangle for closed regions using their boundary points. The vertices of the best-fitted rectangle are computed using a bisection method starting with the upper-estimated rectangle and the under-estimated rectangle. The vertices of the upper- and under-estimated rectangles are directly computed using closed-form solutions by solving for pairs of straight lines. Starting with these two rectangles, we solve for the best-fitted rectangle iteratively using a bisection method. The algorithm stops when the areas of the fitted rectangles remain unchanged during consecutive iterations. Extensive evaluation of our algorithm demonstrates its effectiveness.     

敞口直角容器矩形packing问题的拟人近似算法

众所周知，矩形Ｐａｃｋｉｎｇ问题一般指将众多矩形置入一封闭矩形容器，具有ＮＰ难度。现将封闭容器一端打开变为敞口容器，该文描述了作者在这方面所作的工作。此法改进后，对处理任意多边形极有借鉴意义。     

基于MBR的主方向关系一致性检验

定性的空间推理在地理信息系统、人工智能、数据库及多媒体等领域中的应用越来越引起人们的注意.空间推理的基础理论以及相应算法也在不断地创新和发展.方向关系推理是空间推理研究领域的重要分支,利用区间代数及矩形代数理论,以物体的极小边界盒(minimum bounding rectangle,简称MBR)为模型,提出了一种基于MBR的主方向关系与矩形代数关系相结合的推理方法.利用该方法,可以将矩形代数良好的计算性质应用于空间方向关系推理中,实现了矩形代数与基于MBR主方向关系的相互转换方法、主方向关系合成及求反方法、主方向关系中凸(convex)关系判定方法及方向关系一致性检验算法.     

一种新型的航空图像城区建筑物自动提取方法

提出了一种新的从航空城区图像中自动提取矩形建筑物的方法．该方法基于从航空城区图像中提取的边缘，经过轮廓跟踪，采用Splitting方法提取直线，得出其相应的直线几何图形；针对航空图像的复杂及现有边缘检测算法的不足，提出了一系列直线处理的方法(如直线的分类、排序、合并、调整等)有效地弥补了前述处理的不足；为提高矩形房屋提取的准确率，引入知识定义了几种近似的矩形结构．文章采用几何结构元分析的方法，提取图形中构成矩形的各种基本结构元，再根据结构元合并的准则，将各种基本结构元通过一定的合并算法合并成矩形结构．大量试验结果证明该方法提取矩形房屋的准确率较高，鲁棒性好，运算速度快，具有较强的实际应用价值．     

An iterative merging algorithm for soft rectangle packing and its extension for application of fixed-outline floorplanning of soft modules

We address an important variant of the rectangle packing problem, the soft rectangle packing problem, and explore its problem extension for the fixed-outline floorplanning with soft modules. For the soft rectangle packing problem with zero deadspace, we present an iterative merging packing algorithm that merges all the rectangles into a final composite rectangle in a bottom-up order by iteratively merging two rectangles with the least areas into a composite rectangle, and then shapes and places each pair of sibling rectangles based on the dimensions and position of their composite rectangle in an up-bottom order. We prove that the proposed algorithm can guarantee feasible layout under some conditions, which are weaker as compared with a well-known zero-dead-space packing algorithm. We then provide a deadspace distribution strategy, which can systematically assign deadspace to modules, to extend the iterative merging packing algorithm to deal with soft packing problem with deadspace. For the fixed-outline floorplanning with soft modules problem, we propose an iterative merging packing based hierarchical partitioning algorithm, which adopts a general hierarchical partitioning framework as proposed in the popular PATOMA floorplanner. The framework uses a recursive bipartitioning method to partition the original problem into a set of subproblems, where each subproblem is a soft rectangle packing problem and how to solve the subproblem plays a key role in the final efficiency of the floorplanner. Different from the PATOMA that adopts the zero-dead-space packing algorithm, we adopt our proposed iterative merging packing algorithm for the subproblems. Experiments on the IBM-HB benchmarks show that the proposed packing algorithm is more effective than the zero-dead-space packing algorithm, and experiments on the GSRC benchmarks show that our floorplanning algorithm outperforms three state-of-the-art floorplanners PATOMA, DeFer and UFO, reducing wirelength by 0.2%, 4.0% and 2.3%, respectively.     

Error Estimates of the Finite-Difference Scheme for a One-Dimensional Parabolic Equation with Allowance for the Effect of Initial and Boundary Conditions

Error estimates are obtained for the finite-difference scheme in solving a one-dimensional heat equation. These estimates take into account the effect of initial and boundary conditions. The order of accuracy in a spatiotemporal rectangle is proved to be higher near the rectangle bottom and along the rectangle side on which the Dirichlet boundary condition is specified.     

Corner occupying theorem for the two-dimensional integral rectangle packing problem

This paper proves a corner occupying theorem for the two-dimensional integral rectangle packing problem,stating that if it is possible to orthogonally place n arbitrarily given integral rectangles into an integral rectangular container without overlapping,then we can achieve a feasible packing by successively placing a rectangle onto a bottom-left corner in the container.Based on this theorem,we might develop e?cient heuristic algorithms for solving the integral rectangle packing problem.In fact,as a vague conjecture,this theorem has been implicitly mentioned with different appearances by many people for a long time.     

一种基于三角特征的行人检测算法

基于视频的行人检测技术是计算机视觉领域的重要问题,特征提取是行人检测技术的关键,大量的特征提取方法是基于Papageorgiou提出的矩形特征,但对于行人的边缘特性描述并不准确。针对adaboost算法,分析行人姿态的边缘特性,提出了四种较为有效的三角特征描述人体的外形局部特征,并推导了三角特征计算量与计算耗时的关系。与传统的矩形特征、非对称特征相比,三角特征能更有效的描述行人姿态,提高了行人检测的鲁棒性。     

A new broadband circularly polarized antenna with a single‐layer metasurface

In this article, a new low‐profile broadband circularly polarized antenna with a single‐layer metasurface is designed. The metasurface is composed of 4 × 4 rotated rectangle‐loops. Compared to single rotated rectangle, introducing inner‐cut rectangle slot can increase the design flexibilities by changing this slot size for wider circularly polarized operating bandwidth and reduce the size of the antenna in same frequency. The proposed antenna has the advantages of a wide 3‐dB axial ratio bandwidth from 5.4 to 6.05 GHz and an excellent 10‐dB impedance bandwidth from 5 to 6.05 GHz.     

二维图形最狭长包络矩形的求解原理及方法

最狭长包络矩形是二维图形的一个潜在几何属性,可作为平面外形智能设 计、板料优化排样及图像自动识别的重要依据。目前国内外尚无此课题的专门深入研究。提 出了最狭长包络矩形的概念,将任意二维图形的最狭长包络矩形的求解转化为对其凸包的最 狭长包络矩形的求解。明确给出了过凸包上给定4 个顶点的包络矩形的包络角及长宽比求解 公式,并通过分析包络角及长宽比求解公式之间的关系,证明了凸多边形至少存在一条边与 其最狭长包络矩形的一条边共线。基于该定理,求解并比较与二维图形的凸包的n 条边分别 共线的n 个包络矩形的长宽比,得到了二维图形的最狭长包络矩形。最后用实例验证了定理 和求解方法的正确性和应用效率。     

矩形物体布局的模拟进化算法

介绍了一种称为序列对的表示 Non- Slicing结构平面上矩形块布局的模型 ,并基于该模型提出了一个优化矩形块布局的模拟进化算法     

基于环状矩形分块的纹理图像检索

摘要：提出一种环状矩形分块纹理检索算法。将图像分成环状矩形边缘区域、中心环状矩形区域和核心矩形区域。对最外层边缘区域不予处理,这可明显减少检索时间,且对检索效果影响很少。对每幅图像分别求出中心区域和核心区域基于灰度共生矩阵的能量、对比度和熵所构成的纹理特征向量,两幅图像中心区域纹理特征向量欧式距离和核心区域纹理特征向量的欧式距离之和决定这两幅图像相似性。而环状矩形具有圆形分块的旋转不变性但计算更简单。经实验验证,对一些类别图像有较好的检索效果