The purpose of this research is to develop a new video phone system, which realizes virtual environment as if two speakers stood apart, met in the face through a sheet of transparent glass. In the conventional video phone systems, a camera is set upper or side of a display monitor, because of positional restriction. As a result, the direction of Speaker's view sight does not accords with that of another speaker. On the other hand, using a single eye, we can grasp the 3-dimensional positional relation of objects in sight, by moving its position.
As a first step, we developed the 3-dimensional measurement system, which consists of two video cameras and a liquid crystal display. Using a plaster cast of a head, we evaluated the modeling performance of this system in 1997. As a following step, we performed evaluation of the methods to generate an image of a virtual camera, which was set at the view point of the speaker described above. This image can be automatically generated from the 3-dimensional model and the view point of the speaker.
Last year, we evaluated the overall performances of this system, and made several measures against the problems. At first, in order to eliminate the number of beam scanning for 3-dimensional measurement, the 2-dimensional beam pattern projection method was adopted. Next, the graphic library "OpenGL" and the accelerator were installed to improve the performance of 3-dimensional display.
The major problem of this system is that black hair cannot be measured accurately. It seems that 3-dimensional measurement method using infrared rays is effective to resolve this problem.