摘要
设计与实现了基于微软Kinect摄像机的在线空中手写汉字交互输入与识别实验平台。在研究现有联机手写汉字识别的方法基础上,提出了基于视觉汉字笔画顺序和统计特征的空中手写汉字特征向量提取方法,并利用流形学习方法训练大汉字集的低维流形表示和显式映射函数,结合邻域搜索策略和语义联想可实时高效的进行空中手写汉字识别。该实验平台为空中手写汉字识别方法提供了的客观分析、评价的方式和手段。
This paper designs and realizes an online handwritten Chinese character in the air interactive input and recognition experimental platform based on Microsoft Kinect camera.Based on the research of existing online handwritten Chinese characters recognition methods,the platform utilizes the stroke order and statistic information of characters to learn low-dimensional spaces and explicit projecting functions for big character sets based on a manifold learning method.Besides,combined with neighborhood searching strategy and semantic association,the high recognition accuracy can be achieved in real time.This platform provides an objective analysis and evaluation method for the recognition of Chinese character written in air.
出处
《实验技术与管理》
CAS
北大核心
2016年第7期50-54,共5页
Experimental Technology and Management
基金
国家自然科学基金项目(61402072
61402078
61300083)
关键词
汉字识别
实验平台
人机交互
Chinese character recognition
experimental platform
human-computer interaction