P
197
bit planes in a conventional frame buffer. This combined storage of binary images is called bit-plane stack. Each pixel
(Kx, ky) of the bit-plane stack contains a n-digit bit sequence for the object point recorded here. This bit sequence
unambiguously describes the illumination direction to this object point. In connection with the generated the bit-plane
stack, one can calculate the 3D coordinates of the object points by triangulation if the position of camera and projector
are known.
One of the main advantages of CLA in comparison to other techniques is the speed of the 3D object recording.
This depends on a small number of stripe patterns that have to be projected and recorded by the CCD camera.
The number of projected stripes in the measurement area relates to the resolution of the CCD camera used.
For low precision applications, the recording of the bit-plane stack and the triangulation can be reduced to
simple and fast image processing operations. Depending on the required accuracy and the implemented
algorithms, the complete 3D measurement (up to 250,000 3D coordinates) can be performed in 1 to 10
seconds.
More important than the speed is the robustness of the method regarding to the changing illumination and
different surface reflection attributes. Essentially this criterion is based on the binarizing of the grey level
images by means of a threshold specific to each image point. This threshold is calculated as a mean value of
two images, one dark and one illuminated. So-called shadow areas, which are seen by the CCD camera but not
illuminated by the projector, could be detected by the reference images. A triangulation in the shadow areas is
not possible.
For the projection of the stripes, white-light projectors combined with computer controlled LCD-masks proved
suitable. The available LCD's can be divided into line and grid-mask projectors. In the past, only LCD-grid-
masks were available as a prototype for metrological use. Industrially viable projectors have only recently
become obtainable. These can be controlled in line as well as in column direction. These projectors allow a
variation of the transmission of the lines and columns in several steps. /Wolf 1994/
esa) ANNA
van MIS NS ^
0 AAA 4 te VV. x
n (x) —
Fig. 2 Combination of coded light approach and phase-shifting
The projection of the sine wave intensity-modulated line grid is computer-controlled by variation of
transmission. By means of the phase-shift method and the measured intensity, the phase relation can be
reconstructed. The general principle of this method is to define changes in the phase relation of the projected
grid. The evaluation of the recorded line grid uses the local intensity | (x,y) which is observed with an element of
the CCD-array. It consists of the background intensity H (x,y), the modulation of the stripes M (x,y) and the
phase relation ¢ (x,y) plus a constant phase angle a:
(x, y) 2 H(x, y) [1 * M(x. y) cos(o(x, y) * a)].
IAPRS, Vol. 30, Part 5W1, ISPRS Intercommission Workshop “From Pixels to Sequences”, Zurich, March 22-24 1995