高精度连续人脸轮廓提取

提取连续且高精度的人脸轮廓线是很多图像应用重要的基础步骤.然而很多现有的轮廓提取方法并不能很好地应用在人脸轮廓上.为此提出一种快速可靠的人脸轮廓提取方法,能够在关键点提供初始化后提取到高精度连续人脸轮廓线.其主要步骤是先拟合关键点形成一条初始化曲线,沿其密集采样重叠的矩形区域,将整个人脸轮廓区域划分成很多小的区域;然后在每个局部的矩形区域提取出一条抛物线引导基于梯度的局部人脸轮廓线;最后从很多局部人脸轮廓线中,通过全局融合找到最终的人脸轮廓线.这种交叉验证的机制保证了最后结果的正确性.最后在LFPW和HELEN人脸数据集上进行了实验,结果表明文中方法能有效地提高人脸轮廓提取的精度.     

基于扫描点云数据的零部件边缘轮廓提取技术

针对目前大型零部件边缘轮廓线提取效率低的问题,提出一种基于扫描点云数据的零部件边缘轮廓提取技术.该技术从边缘轮廓线的形成方式出发,通过分析边缘轮廓线垂直截面线的分布特征,定义了模式向量表征屋脊型和折线直线-直线相交型截面线.提取过程主要包含截面线数据的获取、截面线类型的识别和边缘轮廓线数据的提取这3个步骤.为获取边缘轮...     

局部矩不变量轮廓图像角点检测

为解决图像轮廓特征点问题,提出了基于轮廓局部矩不变量的轮廓角点检测方法,通过计算轮廓线支撑区域内中心点两侧轮廓点的不变矩特征,从而提取轮廓曲率函数.实验结果验证了该方法具有抗干扰性好、定位准等优点.     

地面LiDAR数据中建筑轮廓和角点提取

建筑轮廓和角点作为多平台激光雷达数据常用的配准基元,其提取方法正受到越来越多的关注.投影密度法是一种常用的从地面LiDAR数据中提取建筑轮廓和角点的方法,然而以往研究对于直接影响建筑轮廓提取结果的格网密度阈值考虑较少.提出一种轮廓密度估计的方法,能够根据点云实际情况自动准确地计算出格网密度阈值,从而提取较为准确的建筑轮廓格网.在此基础上,利用轮廓线段高程分割和密度延伸的方法对轮廓进行分割和恢复,能够提取完整的建筑轮廓.最后,利用轮廓线段的相交关系获得建筑角点.实验结果表明,本文方法能够有效从地面LiDAR数据中提取建筑轮廓和角点,正确性、完整性和定位精度较高.     

基于曲率匹配的细胞几何形状特征提取方法

细胞轮廓的几何形状是细胞学涂片判读的重要参考,对研究宫颈病变的计算机辅助诊断具有重要意义。针对现有基于形状模板匹配的几何形状识别方法鲁棒性较差的问题,提出了基于曲率匹配的几何形状特征提取方法,通过比较模板轮廓和待识别轮廓的曲率,计算曲率曲线之间的相似度,进而得到细胞轮廓的形状特征,并采用依次旋转轮廓选取最佳匹配的方法来解决轮廓方向不一致的问题,采用以面积等效圆的半径比作为放大比率进行轮廓缩放的方法来解决轮廓大小不一致的问题。通过相关实验证明了该方法所提取的几何形状特征具有尺度不变性和旋转不变性,并与改进Hausdorff距离进行了实验对比,结果表明提取的形状特征能更加准确地识别出细胞轮廓的几何形状。     

非规则三维碎片部分匹配算法研究

针对三维碎片拼合中的碎片匹配问题,在提取碎片轮廓线的基础上,提出了一种快速的碎片匹配算法。该算法首先对轮廓特征点按其邻域曲面片的形状进行分类;其次根据特征点类型标志、曲率及特征段Hausdorff距离对不同轮廓线上的特征段进行相似性度量;最后利用法矢对相似度较高的轮廓段进行可匹配性验证。该算法较适合于用散乱点云表示的三维碎片模型。     

彩色图像中人脸轮廓线提取技术的研究

作为人脸识别的一个重要内容,人脸轮廓线的提取是许多其它艺术影像处理的基础。该文针对复杂背景的彩色图像,提出并实现了一种图像封闭轮廓线提取的方法。该方法利用初始轮廓逐步逼近最佳轮廓,不仅将图像灰度变化的微分信息作为边缘提取的依据,同时以图像轮廓的整体几何信息作为指导,从而在保持边缘检测精度的前提下,一定程度上解决了噪声对轮廓特征提取的影响。     

医学图像中组织轮廓提取方法的改进

在医学图像分割技术的研究中,针对提取图像的清晰度和快速性问题,常用轮廓跟踪算法来提取组织轮廓线,花费的时间较长且分割效果不好.为了提高组织轮廓线的提取速度和分割效果,利用医学图像中组织对称性和灰度特征,提出了双向轮廓跟踪算法.根据分割目标确定相应的检测阈值,采用边缘检测,确定边缘起始点,对组织边缘进行双向跟踪.在检测过程中,设置相应的检测矩阵,减少检测次数.算法能同时沿两个不同的方向进行边缘检测,节省了时间.仿真结果表明,改进的轮廓提取方法对有较好的连续性和对称性的组织轮廓线能进行快速准确地提取,可得到清晰的分割结果,缩短轮廓线的生成时间.     

基于曲率特征的轮廓匹配方法

首先对轮廓曲线进行多边形近似,然后通过Hermite插值曲线求出多边形各顶点的曲率作为特征,最后以Hausdorff距离为准则进行轮廓线匹配。算法充分利用了轮廓线的几何信息,匹配速度快,准确度高,具有一定的旋转不变性。     

人脸边缘自适应搜索算法及仿真研究

人脸轮廓线提取是人脸识别中极为重要的内容,一种可靠、精确的人脸边缘提取算法对于身份鉴定技术等方面具有重要的应用价值。该文基于传统的边缘提取算法提出了一种自适应搜索轮廓线算法,首先基于人脸检测结果确定内外轮廓及搜索路径,然后对于每一条搜索路径提取出真正的轮廓点,最后利用人脸轮廓的平滑性通过曲线拟合完成轮廓线提取。该文以彩色人脸图像库数据为例,快速、准确地得到人脸轮廓线。仿真试验结果表明,该算法能在保持边缘检测精度的情况下,克服了噪声对轮廓特征提取的影响,并且对于姿势变化有一定的鲁棒性,具有一定的应用价值。     

Computing isophotos of surface of revolution and canal surface

Isophote of a surface consists of a loci of surface points whose normal vectors form a constant angle with a given fixed vector. It also serves as a silhouette curve when the constant angle is given as π/2. We present efficient and robust algorithms to compute isophotes of a surface of revolution and a canal surface. For the two kinds of surfaces, each point on the isophote is derived by a closed-form solution. To find each connected component in the isophote, we utilize the feature of surface normals. Both surfaces are decomposed into a set of circles, where the surface normal vectors at points on each circle construct a cone. The vectors which form a constant angle with given fixed vector construct another cone. We compute the parametric range of the connected component of the isophote by computing the parametric values of the surface which derive the tangential intersection of these two cones.     

Minimum distance between a canal surface and a simple surface

The computation of the minimum distance between two objects is an important problem in the applications such as haptic rendering, CAD/CAM, NC verification, robotics and computer graphics. This paper presents a method to compute the minimum distance between a canal surface and a simple surface (i.e. a plane, a natural quadric, or a torus) by finding roots of a function of a single parameter. We utilize the fact that the normals at the closest points between two surfaces are collinear. Given the spine curve C(t), t_min≤t≤t_max, and the radius function r(t) for a canal surface, a point on the spine curve uniquely determines a characteristic circle on the surface. Normals to the canal surface at points on form a cone with a vertex and an axis which is parallel to Then we construct a function of t which expresses the condition that the perpendicular from C(t) to a given simple surface is embedded in the cone of normals to the canal surface at points on K(t). By solving this equation, we find characteristic circles which contain the points of locally minimum distance from the simple surface. Based on these circles, we can compute the minimum distance between given surfaces.     

Bezier surface/surface intersection

The computational requirements and accuracy of two methods for finding the intersection of Bezier surfaces are examined. In both methods, the existence of an intersection curve is confirmed by using the convex hull property of such surfaces. The first method evaluates the intersection by recursive subdivision of two patches with overlapping hulls. The second method detects a point on the intersection curve and then incrementally traces the intersection in the parametric spaces of the two surfaces. With both methods, the intersection of a pair of first-order planar patches must be solved analytically. The intersection is approximated by first-order Bezier patches in the first case and by planar triangles in the second. Overall, the method of incremental tracing is shown to give more accurate results than the method of recursive subdivision     

Detecting the motion of a planar surface by line and surface integrals

A mathematical formulation is presented for detecting the 3D motion of a planar surface from the motion of its perspective image without knowing correspondence of points. The motion is determined explicitly by numerical computation of certain line or surface integrals on the image. The same principle is also used to know the position and orientation of a planar surface fixed in the space by moving the camera or using several appropriately positioned cameras, and no correspondence of points is involved. Some numerical examples are also given.     

Large-scale surface to surface transport for photons and electrons via Monte Carlo

A brief history of Monte Carlo methods is presented, with a differentiation between uses in engineering versus physics, indicating different computational approaches in each field. Particle tracing is presented for photons and then electrons in the presence of a uniform magnetic field. Next, a comprehensive formulation is presented for particle-material interactions, including absorptance, transmittance, and three models of reflectance: diffuse, specular, and semi-specular. Then, a discussion of event steps is presented, indicating that algorithms may be effectively vectorized. Vectorization results are analyzed using a modified version of Amdahl's Law, indicating that over 99% of the “work” of an algorithm may be vectorized.     

Asymmetric surface roughness measurements and meniscus modeling of polysilicon surface micromachined flaps

Polycrystalline silicon (polysilicon) films are primary structural materials for microelectromechanical systems (MEMS). Due to relatively high compliance, large surface-to-volume ratio, and small separation distances, micromachined polysilicon structures are susceptible to surface forces which can result in adhesive failures. Since these forces depend on surface properties especially surface roughness, three types of microhinged flaps were fabricated to characterize their roughness and adhesive meniscus properties. The flaps enabled access to both the top and bottom surfaces of the structural polysilicon layers. Roughness measurements using an atomic force microscope revealed that MEMS surfaces primarily exhibit non-Gaussian surface height distributions, and for the release procedures studied, the bottom surface of the structural layers was significantly smoother and prone to higher adhesion compared to the top surface. A non-symmetric surface roughness model using the Pearson system of frequency curves was coupled with a capillary meniscus adhesion model to analyze the effects of surface roughness parameters (root-mean-square, skewness, and kurtosis), relative humidity, and surface contact angle on the interfacial adhesion energy. Using the measured roughness properties of the flaps, four different surface pairs were simulated and compared to investigate their effects on capillary adhesion. It was found that since the base polysilicon layer (poly0) was rougher than the base silicon nitride and the structural layer on poly0 was also rougher than that on silicon nitride, depositing MEMS devices on poly0 layer rather than directly on silicon nitride will reduce the adhesion energy.     

Pregnant women and working surface height and working surface areas for standing manual work

Physically loading aspects of work may have adverse effects on the health of both the pregnant woman and the unborn child. Improving the fit between the pregnant woman and the workplace layout contributes to minimizing the load associated with given tasks. The aim of this paper is to evaluate two layout aspects for standing manual work, namely working surface height and working surface areas, for the condition of pregnancy. Two approaches were used. (1) The effects of changed body dimensions were evaluated with regard to (a) fit problems while working at a workplace in accordance with common guidelines and (b) the validity of assumptions of these guidelines. (2) The appreciation of relevant aspects of workplace layout at a specific manual task was assessed. Twenty-seven women were examined in pregnant and non-pregnant conditions. The first approach showed that fit problems are likely: guideline working surface height is just (2-7 cm) under the most protruding abdominal point, and areas based on non-pregnant abdominal depth are relatively large in pregnant condition. Further, existing methods to assess working surface areas have various assumptions that are not valid in pregnant condition. The second approach showed that at a specific manual task, women in late pregnancy preferred a considerably lower table height than the common guideline heights. Possibly, abdominal height becomes a relevant design factor with regard to working surface height during pregnancy. The task position on the working surface at which effort started became closer to the table edge due to pregnancy. Both approaches show that common guideline working surface heights for manual work, and working surface areas assessed in non-pregnant condition seem not suitable in pregnant condition.     

Feature-based surface design and machining

A feature-based method for designing and representing functional surfaces such as automobile inner panels that lets a user assemble and present complicated, multifeatured surfaces using known, generally simpler component surfaces and information about feature shape is reviewed. It is shown that, using this method, CAD users can generate numerically controlled (NC) tool paths and use them to automatically machine 3-D surface geometries with various cutting tools such as ball-nose or spherical-end cutters and toroidal or flat-end cutters. The method was tested on models involving simple explicit primary and secondary surfaces as well as more complicated B-spline parametric surfaces. Results indicate that the tool-center-generation algorithm is accurate, robust, and computationally efficient     

Border and surface tracing--theoretical foundations

In this paper we define and study digital manifolds of arbitrary dimension, and provide (in particular)a general theoretical basis for curve or surface tracing in picture analysis. The studies involve properties such as one-dimensionality of digital curves and (n-1)-dimensionality of digital hypersurfaces that makes them discrete analogs of corresponding notions in continuous topology. The presented approach is fully based on the concept of adjacency relation and complements the concept of dimension as common in combinatorial topology. This work appears to be the first one on digital manifolds based ona graph-theoretical definition of dimension. In particular, in the n-dimensional digital space, a digital curve is a one-dimensional object and a digital hypersurface is an (n-1)-dimensional object, as it is in the case of curves and hypersurfaces in the Euclidean space. Relying on the obtained properties of digital hypersurfaces, we propose a uniform approach for studying good pairs defined by separations and obtain a classification of good pairs in arbitrary dimension. We also discuss possible applications of the presented definitions and results.