Development of an Environment of a Distant Lecture with an Automatic Camera Control System

自动摄像控制系统远程讲座环境的开发

基本信息

批准号：
12680215
负责人：
ARITA Daisaku
金额：
$ 2.18万
依托单位：
Kyushu University
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (C)
财政年份：
2000
资助国家：
日本
起止时间：
2000 至 2002
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-12680215/
关键词：
Image processing Supporting system Distant lecture Automatic recording Estimation of lecture Active camera

项目摘要

We have developed the method recording the latest object written on a blackboard. Our method analyzes a lecture scene captured by a steady camera and extracts the latest one by using the subtraction method between two successive frames and the background subtraction method. It is important for our system (ACE) to extract and focus on the explained object so that the system appropriately records the lecture scene for the distant lecture realized by transmitting video and audio of the live lecture. To detect the explained object is so difficult. We assumed that the teacher explained the object as soon as he wrote it. Our system consequently zooms in on the latest object written on the blackboard as the explained object.We have developed a prototype system of ACE and conducted an experiment of applying it to a real 25-minutes lecture on Mathematics for 85 undergraduates. We recorded the lecture scene by ACE and by a steady camera in order to compare them. As the results of our experiment, we confirmed that the scene captured by ACE is good enough for students to read contents on the blackboard. And as the results of our questioner, we made sure that ACE is superior to the steady camera with 1% level of significance.The evaluation of ACE is not high at the question: could you watch something to want. Because students do not always want look at the explained object immediately and ACE does not always extract the accurate latest object. Then we have designed the other method that enabled the students to refer the object they wanted.

我们已经开发了记录在黑板上写的最新对象的方法。我们的方法分析了由稳定相机捕获的讲座场景，并通过使用两个连续帧和背景减法方法之间的减法方法提取最新的摄像头。对于我们的系统（ACE）而言，提取并专注于解释的对象很重要，以便系统适当地记录通过传输现场演讲的视频和音频实现的遥远的演讲场景。检测解释的对象是如此困难。我们假设老师在写作后立即解释了该物体。因此，我们的系统会缩放在黑板上写入的最新对象，如解释的对象。我们开发了ACE的原型系统，并进行了一个实验，将其应用于85个本科生的数学上的真实25分钟的演讲。我们通过ACE和稳定的相机录制了演讲场景，以比较它们。作为我们实验的结果，我们确认ACE捕获的场景足以让学生阅读黑板上的内容。作为我们发问者的结果，我们确保ACE优于稳定的相机，其意义水平为1％。对ACE的评估不高：您可以看一些想要的东西。因为学生并不总是想立即查看解释的对象，而ACE并不总是提取准确的最新对象。然后，我们设计了另一种方法，使学生能够推荐他们想要的对象。