项目名称: 基于Modular积图和最大团的草图形状匹配技术研究
项目编号: No.61305091
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 梁爽
作者单位: 同济大学
项目金额: 25万元
中文摘要: 近年来,随着触摸屏的普及,草图交互技术及其应用已渗透到国民经济的各个行业和国人的日常生活中。在草图识别/检索等很多草图交互系统中,草图形状匹配技术都扮演着基石的角色。草图形状匹配研究的最终目的是希望提出某些度量算法,进而有效、自动地计算两个形状之间的相似度,并希望计算的结果能够和人的视觉感知高度一致。目前的方法对草图形状的匹配能力(尤其是部分匹配能力)还达不到和人的视觉感知高度一致的程度。鉴于此,在本项目中,我们将致力于提出新的高精度的草图形状匹配方法。拟提出的新方法将建立在图模型表示和积图与最大团两个理论基础之上。这是因为:一方面,图模型表示可以有效地处理草图的不确定问题;另一方面,积图与最大团可以灵活地支持部分匹配。我们将尝试把这两个理论有机结合,并有效地引入到草图形状匹配领域中,来构建全新的高精度的草图形状匹配方法。本项目的预期成果将在理论和应用两个方面极大地促进草图交互领域的发展。
中文关键词: 形状匹配;Modular积图;最大团;草图;
英文摘要: With the development of touch-based interfaces, sketch-based applications and interaction become more and more popular in both professional activities and general tasks. Sketch matching plays a significant and fundamental role in various sketch-based applications, such as sketch recognition, sketch retrieval, etc. Actually, the ultimate goal of the research of sketch matching is to design certain computational models that can measure the similarity between two shapes automatically and effectively, in consistency with subjective visual perceptions. Currently, however, the matching ability of the state-of-the-art sketch matching methods, especially partially matching ability, cannot highly correlate with subjective perceptions. To this end, we will lay our focus on devising novel high performance sketch matching methods in this project. Specifically, the newly proposed sketch matching method will be based on graph-based representation and product graph and maximal clique theory, for which there are several justifications. On one hand, graph-based representation is effective in dealing with the characteristics of uncertainty of sketches. On the other hand, the product graph and maximal clique theory is a flexbile and effective way of supporting partial matching in graph matching problems. Thus, in this project, we
英文关键词: Shape Matching;Modular Product Graph;Maximal Clique;Sketch;