草图理解技术研究进展  

Sketch Understanding: Review and Research

作　　者：周若鸿[1] 孙正兴[1] 张莉莎[1] 徐晓刚[1] 

机构地区：[1]南京大学计算机软件新技术国家重点实验室,南京210093

出　　处：《计算机科学》2004年第4期140-146,共7页Computer Science

基　　金：国家自然科学基金(编号:69903006;60373065)

摘　　要：本文概述了草图理解技术(包括草图识别和语义理解)的现状。首先,草图识别包括预处理和图形识别,根据各种图形处理机制,介绍了四种图形识别方法:基于笔划的、基于图元的、基于几何特征的、基于组合图形的。其次,深入分析了几个语义理解方法,一般包括三个环节:语义获取关系分析和高层应用。最后,本文提出了几何模型、用户模型和领域模型。讨论了草图理解中的主要问题并给出初步解决方案。This paper summarizes the state of the art of sketch understanding in aspects of sketch recognition and semantic understanding. Firstly, sketch recognition is divided into preprocessing and graphics recognition, and according to various graphics processing mechanisms, a few graphics recognition methods based on strokes, primitive shapes, geometric features or composite shapes are introduced. Secondly, some typical methods for semantic understanding are analyzed in detail by classifying them into three classes : semantics acquisition, relation analysis and applications. Finally, geometry model, user model and domain model are proposed in this paper, we discuss the key issues in sketch understanding and give preliminary solutions.

关 键 词：草图理解 草图识别 图形识别 图形处理 计算机动画 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]