iX 3D
) grows
igitizers
roduced
e-range
ate 3D-
or 2:5D
'oblems
tionally
ause of
10d for
t space
various
tion of
ith the
ntation
such as
on.
] using
ehr A.,
| object
ruction
erating
image
rtesian
dinates
can be
urface.
ments
s. The
oud. It
Knyaz, Vladimir
allows representing whole surface as uniform triangulation mesh. The accurate united photorealistic texture is generated
for considered 3D models by transformation the set of images.
2 CLOSE-RANGE PHOTOGRAMMETRIC SYSTEM
2.1 Hardware configuration
The presented methods for complex 3D object reconstruction was developed as a part of research digital close-range
photogrammetric system. The processing kernel of system is Pentium personal computer. Also hardware includes:
e Set of high-resolution color and
black and white CCD cameras
of PAL standard for image
acquisition
e Frame grabbers for image
capture
e Set of test fields for purposes of
camera calibration and
orientation
e Controlled structured light
projector
e Controlled turn-table for object
positioning
e Original coded targets for
automated correspondence
problem solution in calibration
and orientation tasks
The view of digital close-range
photogrammetric system is shown in
the Fig. 1.
Figure 1. Digital close-range photogrammetric system
2.2 Software outline
The original software is developed for Windows 95. It supports the complete technology of system calibration,
orientation and object 3D reconstruction.
The software provides:
Image capturing and preprocessing
Automated cameras calibrating (camera internal orientation determining)
Automated exterior and/or relative cameras orientation based on calibrated test fields
Automated coded target recognition and identification with subpixel image coordinates measurement
Feature extraction for automated correspondence problem solution for images obtained in special types of
structured light
Point non-contact 3D measurements in manual and automated (for points marked by coded targets) mode
Automated multiplied points3D measurements (surface scanning) for images structured light
Automated 2.5D surface generation based on standard Delaunay triangulation
Accurate texture mapping (orthophoto generation)
Some methods for automated multiplied non-contact 3D measurements producing a cloud of spatial object coordinate
are developed. They are based on image acquisition in structured light of a stochastic (fractal) form (a), contrast points
form (b) and contrast stripes form (c). The sample images of objects to be reconstructed in corresponding structured
lights are shown in Fig. 2. The first method is based on correlation, the latter use exterior orientation (EO) and epipolar
geometry for 3D reconstruction.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000. 429