next up previous
Next: Engineering achievements Up: Scientific achievements Previous: Publications

Future research plans

Future development of the stereo setup includes research with detailed time-varying texture projected onto the object surface. We beliew that the effect of systematic errors which may be present if the projected texture is the same all the time can be diminished. Moreover, the correspondence matching can be performed in every pixel location leading to a fully measured disparity map.

We have recently started a research project called 'Photorealism in 3-D video digitizing' (3D-IMAGE) and sponsored by TEKES (January 1, 1996 - July 31, 1997). The aim of this project is to implement the 3-D video digitizing procedures into virtual reality modeling tools such as the TargetJR and WorldToolKit. We are going to generate photorealistic solids which are geometrically exact spatial mosaics containing also the radiometric information. The 3-D modeling procedure is based on image information exclusively without any need for geometric reference measures in the object space.

Another research field relates to the question how to process the 3-D data as a function of time when more and more data is acquired. Using the laser-camera profilometer, we are able to scan the object surface from tens or even hundreds of different viewpoints and thus determine the object model more accurately and in greater detail. An error analysis should be performed and used to guide the measuring into areas where prescriebed tolerances are not yet achieved.

In the 3DVIDEO2-project, we considered the problem of registration but no object model was created. Only the measurements were plotted in the same coordinate system. Further research should be carried out in order to find an appropriate object representation for the data in overlap areas. Especially when much more data will be gathered as a function of time, the representation issue becomes important.



Olli T Jokinen
Tue May 28 08:57:45 EET DST 1996