This paper presents a novel approach which continuously detects the first collision between two curved-edge polygons moving under rational motions. Edges of the two polygons in this paper are planar curves, represented as conic splines, i.e. elliptic or parabolic sections. The curved-edge polygons are not confined to be convex and conic sections are only required to be GC0 continuous. Motions of the polygons are modeled by interpolating between control points along motion trajectories. Our algorithm returns the first collision moment and collision position if there is a collision between the two moving polygons and returns no-collision otherwise. Collision condition of the two polygons moving under rational motions is represented as an univariate polynomial of time t. Bernstein form is used to improve the accuracy of solving the high degree polynomial. We also use bounding circles to improve the efficiency of our approach and compare our method with the PIVOT2D method and prove ours to be more accurate and faster.

根据线条画和线条画中笔画的形状特征，运用平面形状演化理论对线条画和笔画的形状进行平滑或增强，并对宽度进行相应的调整，最后得到不同风格的线条画该风格转换不需要任何样本，而且能够控制转换的程度，提高了线条画风格转换的自由度。

给定点云模型上2点，将点云数据沿与驯xyz三坐标轴垂直方向进行单元剖分后，采用Dtlkstra算法求出2点司的最短路径作为初始测地线；然后通过带弧长最短约束的平方距离最小化方法对初始测地线进行迭代优化，计算得到点云模型上给定2点司的一条样条表示的精确测地线文中算法只需局部拟合抛物曲面，无需对点云模型进行三角化或曲面重建，适合大规模点云数据模型上测地线的计算。

为了能采用较少种油墨印制出色彩鲜艳的连续调图像，提出了一种面向专色印制的油墨定制及连续调图像分色算法。连续调图像是指存在着色调、亮度与饱和度的连续变化的真彩色图像。该算法首先根据图像的颜色属性，建立油墨定制与分色的约束最优化模型，然后采用迭代法来完成油墨定制及图像分色。实践证明，使用该算法能够以较少种类的油墨，在不允许叠印的前提下，印制出色彩鲜明的连续调图像。该算法还可广泛应用于印染、陶瓷印花等行业。

提出了一种利用线条画作为样本对任意线条画进行风格转换和定制的方法，该方法不仅可以将样本图风格转移到目标图，还可以方便地控制目标图与样本图之间风格的相似程度。该方法主要采用基于细分笔划的风格相似性比较算法，可在多层次上把目标图转换为样本图风格，不仅如此，还通过定义风格的插值操作实现对风格的定制。已有方法相比，文中方法只需少量的样本笔划即可进行风格转换，并且通过风格定制，使风格转换具较强的交互可控性。文章最后通过实例证明了方法的有效性。

提出了一个构造随机聚合网屏的新算法。该算法首先构造一个大抖动矩阵，随机遍历抖动矩阵的元素，在自由元索处放置选定半径的离散圆盘，获得网点的中心，这样构造出一个由数千个随机放置的网点组成的大抖动矩阵，然后对网点中心进行Delaunay三角化，计算出每个网点的最大表面。再把每个网点划分成256份等密度区域。得到一个初步的值矩阵。对这个圈值矩阵进行平滑处理，可得到最后的抖动矩阵，通过改变盘子的半径，司以使网点的大小能适应特定的印刷设备。实验证明，使用这种方法产生的随机网屏对高分辨率的印刷设备效果显著，1200dpi的打印机用半径r=8产生的随机网屏，相当于150lpi的传统网屏。

提出一种利用带约束规则的变形图案生成各种笔划效果的方法。将笔划定义成一个图案模板（称为笔划模板），并运用基于路径的笔划模板变形算法来产生笔划。引入了三种变形约束规则：位置约束、形状约束和重复性约束。通过这些约束规则，可以控制笔划的变形效果，达到“智能变形”的目的。作为一种非真实感图形绘制技术，文 中方法可以用于艺术创作、印染以及陶瓷花纸等工业的图案设计领域。

Levin's method produces a parameterization of the intersection curve of two quadrics in the form p (u)=a (u)±d (u)s (u), where a (u) and d (u) are vector valued polynomials, and s (u) is a quartic polynomial. This method, however, is incapable of classifying the morphology of the intersection curve, in terms of reducibility, singularity, and the number of connected components, which is critical structural information required by solid modeling applications. We study the theoretical foundation of Levin's method, as well as the parameterization p (u) it produces. The following contributions are presented in this paper: (1) It is shown how the roots of s (u) can be used to classify the morphology of an irreducible intersection curve of two quadric surfaces. (2) An enhanced version of Levin's method is proposed that, besides classifying the morphology of the intersection curve of two quadrics, produces a rational parameterization of the curve if the curve is singular. (3) A simple geometric proof is given for the existence of a real ruled quadric in any quadric pencil, which is the key result on which Levin's method is based. These results enhance the capability of Levin's method in processing the intersection curve of two general quadrics within its own self-contained framework.