物体轮廓线的多边形拟合

本文讨论了在给定数字轮廓线顶点数目为N的情况下，从其上选择k个点来构造拟和多边形来近似原图，使得物体轮廓线的形状丢失最小。本文的贡献在于1）本文的轮廓线拟和是以物体轮廓线的形状的信息丢失最小为目标的。而传统的方法是以拟和多边形和原图的面积差距最小。2）本文认为每一个点具有一定的形状信息，然后建立了一个以丢失的形状信息最小的0—1规划模型：3）使用了两种方：去对轮廓线进行了多边形拟合，并把两个结果进行了比较。     

三维重构中轮廓线间的连接和光滑问题的处理

在使用面绘制算法重构三维实体模型时,由于原始数据稀疏,需要通过一定的方法对填充在相邻轮廓线间的三角形或多边形进行拟和,以达到光滑的效果。本文先按照最小内角最大准则进行Delaunay三角剖分,当可选三角形的最小内角相等时再运用最短路径法在相邻轮廓线间构造三角形,然后再在三角格网上构造Bezier三角曲面,不仅使构造出来的格网具有较好形状,又提高了表面的光滑程度和重构的精度。     

一种双向层间轮廓线线性插值方法

CT和MRI图像断层之间的距离通常大于断层内部间的像素的距离,为了实现断层图像层间的合理过度,需要对轮廓线进行线性插值。文中提出了一种双向的断层轮廓线线性插值方法。该方法首先对轮廓多边形进行逼近,然后从两个方向寻找最佳匹配点对,最后进行线性插值。该方法易于实现,执行速度快。实验结果表明这种方法不仅对形状相似的轮廓线插值效果好,对形状差异大的轮廓线也能得到令人满意的结果。     

一种双向层间轮廓线线性插值方法

CT和MRI图像断层之间的距离通常大于断层内部间的像素的距离,为了实现断层图像层间的合理过度,需要对轮廓线进行线性插值.文中提出了一种双向的断层轮廓线线性插值方法.该方法首先对轮廓多边形进行逼近,然后从两个方向寻找最佳匹配点对,最后进行线性插值.该方法易于实现,执行速度快.实验结果表明这种方法不仅对形状相似的轮廓线插值效果好,对形状差异大的轮廓线也能得到令人满意的结果.     

基于质心距离增量矩阵的曲线描述和匹配

图像中物体轮廓线的描述和辨识是图像理解的一个核心问题,提出了一种新的快速的轮廓线描述和匹配方法。该方法不是把曲线中的每一点都进行了描述和匹配,而是按照最小面积法则对曲线进行采样。用采样点与质心的距离增量为每一个采样点赋予一个身份向量,通过所有采样点的身份向量来构造曲线的质心距离增量矩阵以描述曲线,最后利用两条曲线的质心距离增量矩阵的差异度来度量它们的相似性。实验证明该描述方法不仅具有很好的唯一性,而且相比目前已有算法计算量大大减少。该方法的一个实际应用在于将一个地区的SAR图像与地图相匹配。     

基于轮廓分割的形状匹配新方法

提出了基于轮廓分割的形状描述方法,用来进行形状匹配.首先,使用离散曲线演化将形状轮廓简化成具有N个顶点的多边形,并通过去除其中的凹点得到多边形凸点的集合;然后,对由凸点形成的轮廓片段提取角度和曲率特征,得到形状描述子;最后通过匈牙利算法进行轮廓片段之间的距离矩阵的最小代价和计算,从而得到两个形状之间的相似度.基于MPEG-7图像库上的实验证明了本文方法能很好地反映形状的特征,具有较好的匹配效果.     

基于图形硬件的光滑线条绘制(英文)

线条能够以相当少的可视信息来有效地表示一个三维模型的形状.利用图形硬件的高度并行处理能力,文中提出了一种新的基于物体空间的线条绘制方法.对于一个光滑物体形状,新型的Suggestive轮廓线能够结合侧影轮廓线来生成非常好的视点相关的线条效果.并且,最新被引入到Direet3D 10流水线的几何着色器能处理三角形图元数据及输出零个或多个线条图元.文中提出了一种并行化的线条绘制方法,能够直接在几何着色器中抽取三维的特征线条.该方法能根据视点与模型的距离自动地选取合适比例的线条进行绘制,绘制的线条平滑,绘制速度快.在文章最后给出的许多实验示例证实了该方法的效果.     

基于多幅实拍照片为真实景物模型添加纹理

利用实拍照片为基于真实景物创建的几何模型添加纹理的方法正在受到广泛的关注,实拍照片与几何模型的配准是这项技术的关键.以往方法采用3D-2D特征点匹配或侧影轮廓线匹配的方法进行配准,因此对空间物体的表面特征或轮廓线形状有特殊的要求.提出了一种新的配准方法来解决这一问题,由于采用了基于图像重建的采样点模型与已知几何模型在空间中匹配的方法实现配准,因而充分利用了物体几何形状本身的拓扑和曲率等信息,并可以一次性地实现所有图像与空间物体的配准.实验结果表明,该方法可以解决一部分用以往的方法尚无法处理的实际问题,且在重建空间采样点分布较为合理的情形下,纹理映射效果非常理想.     

基于实测轮廓线的岔道口三维实体建模方法

在数字矿山系统中,为使错综复杂的地下巷道重建效果能达到矿山的验收标准,根据实测巷道交岔口轮廓线,采用分区(分块)建模思想,提出一种基于实测轮廓线的岔道口三维实体建模方法。将实测的岔道口轮廓数据按其特点分为顶底板和侧面轮廓线点集,采用凸包算法构造顶底板的简单多边形,并对其进行三角化。根据顶底板多边形顶点的先后顺序关系,建立侧面轮廓线的相邻关系,并采用连线框算法对相邻轮廓线进行三角化。实验结果表明,该方法简单有效,可解决任意断面形状、多个岔道口的建模问题,并且已经应用于DIMINE软件的测量插件中,效果良好。     

基于轮廓线匹配的二维重迭物体的识别

本文提出一种基于轮廓线匹配的二维重迭物体的识别算法,并采用一种点模式匹配的快 速算法进行特征点序列的匹配,通过引进形状移位数的概念和利用问题中的各种约束条件,提 高了物体识别速度.     

Real Time Fitting of Hand-Sketched Pressure Brushstrokes

A method is described for fitting the outline of hand-sketched pressure brushstrokes with Bézier curves. It combines the brush-trajectory model, in which a stroke is generated by dragging a brush along a given trajectory, with a fast curve fitting algorithm. The method has been implemented for a vector-based drawing program in which the user draws with a cordless pressure-sensitive stylus on a digitizing tablet. From the trajectory followed by the stylus, its associated pressure data, and a specified brush, a stroke of variable width is computed and displayed in real time. First, the digitized trajectory is fitted, thus removing noise. Then, from polygonal approximations of the fitted trajectory and the brush outline, a polygonal approximation of the stroke outline is computed. Working with polygonal approximations reduces computations to simple geometric operations and greatly simplifies the treatment of dynamic, pressure-controlled brushes. Last, the polygonal approximation of the stroke outline is fitted. The result is a closed piecewise Bézier curve approximating the brushstroke outline to within an arbitrary error tolerance. Several examples of hand-sketched drawings realized with this method are presented.     

How to merge hidden arcs and then not draw them

The procedures described here apply to a process which is formulated as a function of a continuous scalar parameter over a finite range. The problem considered is how to generate the function with the parameter being incremented in a series of discrete steps, when a series of subranges of parameter values are to be omitted.

The application described is the function of drawing of ellipsoid outlines when parts of the outlines are obscured by other ellipsoids. The scalar parameter is an auxiliary angle running around the outline from 0 to 2π which is incremented in finite steps generating a series of points around the outline. These are connected by chords giving a polygonal approximation to the outline.

It is shown that it is desirable to do the drawing using an ordered consolidated list of parameter values at the starts and ends of the hidden arcs. Algorithms for generating and using this list are presented and discussed.

Generalisations of the procedures are discussed.     

Polygonal approximation of planar curves in the L 1 norm

An algorithm for obtaining a polygonal approximation in the L ₁, norm of a plane curve of arbitrary shape is presented. The L ₁ error norm in any segment is not to exceed a pre-assigned value. The given curve is first digitized and the algorithm is then applied to the discrete points. The algorithm uses linear programming techniques which makes it efficient and fast. Numerical results and comments are given.     

基于遗传算法的多边形逼近3D数字曲线

首先对3D数字曲线进行简单的数据压缩．通过对该曲线上的点列进行二进制编码定义来表示数字曲线的染色体．二进制串中的每一个位称为基因，每一个逼近多边形和染色体形成1-1映射．目标函数使给定曲线和逼近多边形之间的均方差最小．构造了解决该问题的选择、交叉、变异三个算子．所得最优染色体中基因值为1的基因对应数字曲线的分界点．实验结果表明，该方法能够得到精确的逼近结果．     

Model based foot shape classification using 2D foot outlines

This study introduces a novel technique to identify foot outline characteristics and to classify feet into groups using turning functions and clustering techniques so that shape can complement anthropometry in producing good fitting shoes. The digital 3D foot scans, obtained from 50 Hong Kong Chinese subjects (25 males and 25 females) were processed to generate the foot outlines at heights of 2 mm and 40 mm. The outlines were represented as turning functions and the similarity among shapes was determined using average linkage clustering. The results show that there are two distinct shape groups for the 40 mm foot outlines on both medial and lateral sides of the foot. The presence (46%) or absence (54%) of a medial bulge characterizes the medial side, while the two shape groups on the lateral side are mainly due to the lateral concavity in the mid-foot region. The group with a lateral concavity consists of more females (68%) and thus lateral side of foot outline appears to be gender related. Furthermore, the medial and lateral side clusters are not related to each other. The medial side shape from the 2 mm foot outline is a good indicator of fallen arches. Based on the analyses, four types of feet were identified: feet with (1) lateral concavity and a medial bulge, (2) a medial bulge and no lateral concavity, (3) lateral concavity and no medial bulge and (4) lateral concavity and a medial bulge. These shape differences can be useful in the design of shoe lasts and in the manufacture of compatible footwear so that trial and error fitting can be minimized.     

An approach to averaging digitized plantagram curves

The averaging of outline shapes of the human foot for the purposes of determining information concerning foot shape and dimension within the context of comfort of fit of sport shoes is approached as a mathematical problem. An outline of the human footprint is obtained by standard procedures and the curvature is traced with a Hewlett Packard Digitizer. The paper describes the determination of an alignment axis, the identification of two ray centres and the division of the total curve into two overlapping arcs. Each arc is divided by equiangular rays which intersect chords between digitized points describing the arc. The radial distance of each ray is averaged within groups of foot lengths which vary by ± 2·25 mm (approximately equal to 1/2 shoe size). The method has been used to determine average plantar curves in a study of 1197 North American males (Hawes and Sovak 1993).     

Optimal polygonal approximation of digitized curves using the sum of square deviations criterion

We address the problem of finding an optimal polygonal approximation of digitized curves. Several optimal algorithms have already been proposed to determine the minimum number of points that define a polygonal approximation of the curve, with respect to a criterion error. We present a new algorithm with reasonable complexity to determine the optimal polygonal approximation using the mean square error norm. The idea is to estimate the remaining number of segments and to integrate the cost in the A^* algorithm. The solution is optimal in the minimum number of segments.     

A new algorithm for dominant points detection and polygonization of digital curves

A new algorithm for detecting dominant points and polygonal approximation of digitized closed curves is presented. It uses an optimal criterion for determining the region-of-support of each boundary point, and a new mechanism for selecting the dominant points. The algorithm does not require an input parameter, and can handle shapes that contain features of multiple sizes efficiently. In addition, the approximating polygon preserves the symmetry of the shape.     

Cubic Bézier approximation of a digitized curve

In this paper we present an efficient technique for piecewise cubic Bézier approximation of digitized curve. An adaptive breakpoint detection method divides a digital curve into a number of segments and each segment is approximated by a cubic Bézier curve so that the approximation error is minimized. Initial approximated Bézier control points for each of the segments are obtained by interpolation technique i.e. by the reverse recursion of De Castaljau's algorithm. Two methods, two-dimensional logarithmic search algorithm (TDLSA) and an evolutionary search algorithm (ESA), are introduced to find the best-fit Bézier control points from the approximate interpolated control points. ESA based refinement is proved to be better experimentally. Experimental results show that Bézier approximation of a digitized curve is much more accurate and uses less number of points compared to other approximation techniques.     

A bivariate autoregressive technique for analysis andclassification of planar shapes

A bivariate autoregressive model is introduced for the analysis and classification of closed planar shapes. The boundary coordinate sequence of a digitized binary image is sampled to produce a polygonal approximation to an object's shape. This circular sample sequence is then represented by a vector autoregressive difference equation which models the individual Cartesian coordinate sequences as well as coordinate interdependencies. Several classification features which are functions or transformations of the estimated coefficient matrices and the associated residual error covariance matrices are developed. These features are shown to be invariant to object transformations such as translation, rotation, and scaling. Laboratory experiments involving object sets representative of industrial shapes are presented. Superior classification results are demonstrated