一种基于图匹配的复杂草图识别方法  

A Method of Graph-Based Composite Sketchy Graphics Recognition

作　　者：张莉莎[1] 孙正兴[1] 周若鸿[1] 徐晓刚[1] 

机构地区：[1]南京大学计算机科学与技术系多媒体技术研究所,南京210093

出　　处：《计算机科学》2004年第4期147-150,共4页Computer Science

基　　金：国家自然科学基金(编号:69903006;60373065)

摘　　要：复杂草图识别是手绘草图输入中的一个困难任务。现存的草图识别方法强调的是图形对象简单,但这不适合具有不同复杂性的复杂草图的识别。本文对具有不同复杂性的复杂图形对象提出一个基于图形的统一表示法,文中根据不同信息粒度将复杂图形分别转化为空间关系图(SRG)。文中提出了一个约束的部分枚举,以减小识别复杂草图时匹配SRG的状态空间。实验结果显示,我们的方法可适用于具有不同复杂度的各种复杂图形对象的识别。Composite sketchy graphics recognition is a difficult task in sketch-based graphics input. Existing sketchy graphics recognition ways have emphasized on simple graphic objects and have not suited for recognition of composite graphic objects with different complexity. This paper presents a graph-based uniform representation for composite graphic objects with different complexity, which, concerts them respectively into Spatial Relation Graph (SRGs)in different information granularity. Accordingly, a Constrained Partial Permutation algorithm is proposed to. reduce state space in matching SRGs for recqgnizing composite sketchy graphics by predicting the directly next state of the current state and neglect invalid states in order- Experiments show that the proposed method can be adapted for recognition of various composite graphic objects with different complexity.

关 键 词：草图识别 图匹配 图形识别 匹配算法 计算机 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]