All the techniques developed in the 3DVIDEO and 3DVIDEO2 -projects have been implemented, verified, and demonstrated in laboratory conditions and most of them tested also in real situations.
In the 3DVIDEO-project, the stereo setup was built and the practical procedure for automated calibration (developed earlier) installed. The stereo setup consists of two CCD video cameras and an actively controlled feature projector. The cameras are mounted convergently on a common basement. The setup is internally adjustable according to different object distances by changing the distance or the convergence angle of the cameras. For each internal setup geometry, the mutual coordination is automatically calibrated by means of a set of corresponding point observations. The calibration is based on projective equations and produces a rectified image pair which comforms to the normal case of stereo photogrammetry. Due to the normal case geometry, the corresponding object elements are lying pairwise exactly along the one and same horizontal line of both images. The image distance between each pair of corresponding image elements is called disparity and relates to the 3-D geometry of the object.
A set of photographic slides produced synthetically are used to project featureness on the object surface. The digitizing of one view consists of several subsequent phases of single digitizings. During each single digitizing both images are scanned in order to collect all identified correspondences. The disparity information is stored into a disparity map. The correspondences are measured automatically and controlled by means of the active feature projection positioned in the mid of the camera base. The measuring system is capable of controlling the precision of the 3-D matching procedure and projecting supplementary featureness and detailness on the object surface during the digitizing in order to fulfill the needs set for the completeness and accuracy of the object model under reconstruction.
During the 3DVIDEO-project, we also verified the digitizing based on
a single camera and laser. We had a HeNe laser with a cylindrical lens
to illuminate a stripe on the object surface and the laser stripe was
recorded by a Pulnix TM-6CN or TM-560 (
pixels)
CCD video camera with a band-pass filter. In laboratory conditions,
the object movement was carried out manually in a bench vice.
In airborne applications, the laser-camera profilometer was installed
in a helicopter for the scanning.
In the 3DVIDEO2-project, we built a computer-controlled driveway
for very accurate (location error 
m) stepwise translational movement
of the object perpendicular to the plane defined by the laser sheet
and synchronized with the image recording. In this way, we are able to scan
40 000 points in five minutes from the object surface visible to the
profilometer in close-range applications.
In order to measure the object from all directions, the object is rotated
and translated to different positions and for every position, a new
profile map is acquired.
Another outcome of the 3DVIDEO2-project lies in the design and simulation of video digitizing using 3-D CAD-models [Ø ]. This research was partly funded by the 3DVIDEO2-project and partly by another TEKES project called 3D2000 (Sensor fusion and measurement model in 3-D vision). We added more functions for photogrammetric measurement to the tool for measurement model design (MMD) developed mainly in VTT Electronics. The functions provide information about precision and reliability of the measurements achievable with a given setup of cameras. The computed precision measures like standard errors of the unknown coordinates, error ellipses in the coordinate planes, error ellipsoids and mean radial spherical error are visualized at every planned measurement point in the CAD-environment.