A development on human interfaces aided for remote education

本课题的目标是基于人机界面辅助远程教育的基础开发，利用大学课堂上常用的黑板或白板来制作远程教育的内容。本项目包括以下两个基础研究。(1)远程教学摄像机视角的自动确定我们正在研究一种新的摄像机视角自动确定算法，该算法通过互联网将讲课的动态图像传输给远程学生，该算法使用的是由计算机控制的活动摄像机，无需摄像师。在本课题中，我们提出了一种基于模糊推理和运动图像处理的主动手势识别算法，并通过一些讲师验证了该算法的有效性。(2)手写字符识别和知识库的自动生成我们讨论了黑板或白板上手写字符的分割和识别算法，以及对打印文档的表格形式等文档结构的理解算法，以制作讲座上的知识库。我们提出了一种新的触摸手写字符分割算法和一种新的表结构分析方法来翻译数字（XML）文档。

The goal of this project is a basic development on human interfaces aided for remote education, and we are using the usual lectures using the blackboard or whiteboard in university to make the contents of remote education. The project includes two following basic researches.(1) Automatic determination of the camera view to remote educationWe are considering the new algorithm for the automatic determination of the camera view to transmit the moving images of the lecture to remote students via the Internet, and it uses active camera controlled by computer without cameraman. In this project, we proposed the algorithm of active gesture recognition by using fuzzy reasoning and image processing by the moving images, and confirm the usefulness by some lecturers.(2) Handwritten character recognition and the automatic generation of knowledge databaseWe discussed the segmentation and recognition algorithm of handwritten characters on blackboard or whiteboard, and the understanding algorithm of the document structure such as the table form for printed document to make the knowledge database on the lecture. We proposed a new segmentation algorithm for touching handwritten characters and a new analysis method for table structures to translate digital (XML) documents.