快速凸包计算实现及其应用-信号与信息处理专业论文.docxVIP

摘要I 摘要 I 摘要 凸包是计算几何中最普遍、最基本的一种结构，广泛的应用于模式识别、碰 撞检波、冶金术、城市规划、制图学、图像处理、数值积分、统计学。凸包的研 究，从它的概念的提出至今一直是国内外计算几何领域研究的热点之一。在过去 的几十年中，凸包算法取得了很大的进展。在二维平面上有 Graham 扫描法、Jarvis 步进法和分治算法等，三维空间上有卷包裹法和随机增量算法以及 Rd 空间上的快 速凸包算法。 本文首先介绍了凸包的性质和凸包计算的复杂度，并在此基础上对平面点集 上的 Graham 扫描法、Jarvis 步进法和分治算法以及三维空间上的卷包裹法和随机 增量算法进行了分析。并重点讨论了快速凸包算法。该算法是在随机增量算法的 基础上发展的。主要的改进在于：随机增量算法中加入点集中任一点的操作模式 被改进为寻找当前凸包外最远的点进行加入。 而且，也采用了一些数据结构上的 改进去加速凸包计算算法的速度和效率。这部分的主要贡献在于凸包计算算法的 快速实现。 另外，作为凸包算法的应用，为了解决两类不同特征的训练数据的分类问题， 区域收缩算法被提出用来确定判决域。实验证明在两类训练样本不可分的情况下， 提出的训练算法是有效的。 关键词：凸集 凸包 快速凸包算法 随机增量 ABS ABSTRACT III ABSTRACT The convex hull, as one of the most common and fundamental structure in computational geometry, is widely used in many applications, for example, pattern recognition, collision detection, metallurgical technology, city planning, cartography, image processing, numerical integration and statistics etc.. It has always been one of the most popular topics in the field of computational geometry since the concept of the convex hull was proposed. In the past decade years, there has been great rapid progress in the algorithms of convex hulls computation: Graham algorithm, Jarvis’ step-by-step algorithm and divide and conquer in 2-D and gift-wrapping, randomized incremental algorithms in 3-D space and the Quickhull algorithms in Rd. In the thesis, the properties of the convex hull and its computational complexity are firstly introduced. The analysis is made on several algorithms, for instance, the Graham algorithm, Jarvis’ step-by-step algorithm, and the divide-conquer algorithm for a set of planar points, as well as the gift-wrapping, randomized incremental algorithms in the three-dimensional space. Particularly, the Quickhull algorithm for convex hull computation is paid more attention. It is an improvement of the randomized incremental algorithm. The furthest point of an outside set is selected to be added into the convex hull, instead of a random point as in the randomized incremental algorithm. Also, some