复杂几何对象的特征提取及分类和检索研究

项目名称： 复杂几何对象的特征提取及分类和检索研究

项目编号： No.61272026

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 宋瑞霞

作者单位： 北方工业大学

项目金额： 60万元

中文摘要： 针对平面或立体的由多个几何对象构成的复杂群组，研究：1.在对复杂群组作整体数学表达时，怎样消除通常不可避免的Gibbs现象？2.面对大量的复杂群组，怎样给出群组间差别的量化标准？3.复杂群组的信息处理将面对巨量数据，怎样设计有效的实用算法？4.怎样检验前述3个问题的研究成果的实效？这4个问题在对象分类、模式识别等领域有重要的实用价值。本项研究的基本思想基于非连续正交完备函数V-系统的理论；V-系统的基础理论由申请人提出、并在上一个国家自然基金项目中完成，此为已有基础；本项目将聚焦V-系统的特殊性质进行深入研究，给出多变量V-系统的构造性定义与算法，紧密结合商标检索和模型重用的实际背景，提供具体实践检验的成功例证。本项目的特色首先表现在对复杂群组整体特征的关注，创新在于扬弃连续的正交函数系、而选用非连续的V-系统，解决复杂群组含有间断特性的瓶颈难点问题，并强调以商标检索来检验理论成果的实效。

中文关键词： 正交函数；V-系统；特征提取；几何群组；形状检索

英文摘要： For a complex object group comprising a plurality of 2D or 3D geometric objects, this proposal aims at conducting the following studies: (1) when a complex group is expressed as a single mathematical expression, how to eliminate the effects of the Gibbs phenomenon. (2) Given a large number of complex object groups, how to quantify the difference between them. (3) Since huge amonts of data will be involved when we proccess imformation of complex obeject groups, it is necessary to devlope effective and practical algorithm. How to design such kind of algorithm? (4) How to evaluate the actual effect of the research results of the previous three issues. The research results of these 4 problems have important practical value in object classification, pattern recognition and other fields. This proposal employs the discontinuous orthogonal complete function system, V-system, as a theoretical tool. The theorys of the V-system was proposed by the applicant, which was completed in the last National Natural Science Fund Project. Based on it, this proposal will focus on some distinctive characteristics of the V-system and continue in-depth study. First, the constructive definition of the multi-variable V-system will be proposed and corresponding algorithms will be developed. Then, successful experiment results will be provid

英文关键词： Orthogonal functions；V-system；Feature；Geometric group；Shape retrieval