The basis for this project lies in our earlier research on 3-D video digitizing and subsequent data processing during 1992 - 1996.
In 1992 - 1993, we developed the basic measuring techniques for 3-D video digitizing of objects and space extending the videogrammetric measuring from pointwise to profilewise and surfacewise measurements [9-17,48,49]. The techniques were realized using either a movable stereo head and a feature projector, or a laser-camera profilometer scanning the object surface stepwise. The techniques were implemented, verified, and demonstrated in laboratory and real conditions.
In 1994 - 1995, we developed techniques for subsequent data processing [1,29-31] including methods for outlier removal, methods for the registration of multiple 3-D data sets acquired from different viewpoints [32,33], and tools for the design and simulation of video digitizing by using CAD-models. We built also a computer-controlled driveway with very accurate translational object movement to automate the scanning in the laser-camera profilometer.
The research on photorealistic models was started within a Master's thesis [25] in 1995. The research has been continued in a project ``Photorealism in 3-D video digitizing'' (3D-IMAGE; 1.3.96 - 30.9.97) financiated by TEKES and HUT. In the 3D-IMAGE project, we will reconstruct photorealistic models with the help of the tools of a virtual reality software called ``WorldToolKit'' (by Sense8 Corporation, USA). The basic 3-D video digitizing procedures will also be implemented into another 3-D modeling software called ``TargetJR'' (by General Electric, Inc., USA). Our approach in the 3D-IMAGE project is a practical one involving quite a lot manual work during photorealistic modeling.
Quite recently, we have linked all the microprocessors in the computers
of our laboratory (15 PC 386's or 486's and a Pentium PC) to
process in parallel. The realization is based on the MultiFunctional
Distributed Shared Memory Architecture [3-7] in a systolic machine
allowing many stages of processing to take place simultaneously. In that
environment, we can process the tasks of videogrammetry in near real time.
We have also under construction a multi-camera setup where four cameras
are installed convergently into the corners of a rectangular cage
( 
meters). The cameras can be moved along the vertical
edges of the cage to able a fast scanning of an object inside the cage.
We have thus created the scientific and technical opportunities for developing the 3-D video digitizing techniques further by taking the time variable into consideration. We may gather more and more data at video frequency from a fixed viewpoint and we are able to scan an object surface from tens or even hundreds of different viewpoints and thus determine the object model more completely and in greater detail. Estimating the accuracy of the object model gives also information for the guidance of the measuring into areas where prescriebed tolerances have not yet been achieved. Within this project, we consider the fundamental issues related to the processing of the vast amount of data that the video cameras produce as a function of time. The methods will then be applied to photorealistic modeling in the related 3D-IMAGE project.