具有实时反馈的草图交互识别方法  

An Interactive Sketch Recognition Approach with Real-time Feedback

作　　者：李俊峰[1] 张习文[1] 敖翔[1] 戴国忠[1] 

机构地区：[1]中国科学院软件研究所人机交互技术与智能信息处理实验室,北京100080

出　　处：《计算机辅助设计与图形学学报》2005年第11期2453-2458,共6页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(60273024;60033020);国家重点基础研究发展规划项目(2002CB312103)

摘　　要：针对同步在线草图识别算法中的效率和应用范围问题,提出基于增量式意图提取的识别算法.算法通过定义滞后窗口,采用增量式意图提取的方式理解用户的勾画意图,进而根据当前信息修正以前生成的意图段落,使得识别结果和用户的勾画意图保持一致.实验证明,该算法能够准确、实时地识别用户输入的多种图形.In order to promote efficiency and extend applicability of the online synchronous sketch recognition, a sketch recognition algorithm based on incremental intention extraction is presented. Through defining a lag window, the algorithm understands the sketch intention of users on the base of incremental intention extraction. Moreover, the algorithm can update the existing intention sections according to the latest information such that the recognition results are in line with the sketch intention of users Experiments show the algorithm can recognize various kinds of sketches accurately in real time.

关 键 词：同步在线识别 增量式意图提取 草图识别 滞后窗口 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]