The compensation principle is to reduce the gap of the plate
observations, [ r r, ] — 0. The matrix is diagonal in shape and
its resolution goes through the RNM (Reduced Normal Matrix
[10]. Concerning the external data, a condition equation on the
distances reduced to the scale of the plate. This fusion of
photogrammetric precision and the potential of interpretation of
the object characteristics is a new approach associating
geometry and radiometry [11].
With the evolution of the photogrammetric systems with video
images, combination with external data in the total treatment
will allow a better mastery of space reference and a more
precise assessment on the object [12].
Algorithm, figure (5)
= To acquire perspective i and i+1;
> Pre-treatments (x,y), («', y"), oblique;
> DLT treatment - calculation of distortion — interpolation;
> Storage of parameters (14) ;
=> Repeat 1,2,3,4 until i+1 = NR beams
=> Integration of other measurements (distances, co-ordinates,
etc..)
=> Adjustment of parameters (beams) - Rigorous method;
— dimensional control.
Sensor
>) Computer
Perenective |
Treatments
Perspective 2
Perspective N
Selection and capture of
points
Calculation of the
re transformation parameters
Calculation of residue,
ES standard deviation , and
verification
Calculation of distortion
Interpolation
md
Calculation of the new points
Figure 5 - Software architecture
4 - Application
4-1- Test Object
Our example concerns an experiment carried out in the National
Center of Space Techniques (CNTS) on an industrial object (a
Mazda pick up truck), figure (6)) The three-dimensional
reference observations are carried out using an electronic
tachymeter (ZEISS ELTA 46), and the snapshots using a simple
camera (Minolta) with a 45 mm lens. Th overlap of the two
beams was 8096. The calculation of the ground co-ordinates
(object) and the three dimensional extraction includes all of the
image observations of the points. The latter were carried out on
a C100 ZEISS Planicomp comparator on the one hand, coupled
with an on screen observation on the other hand. This double
task was necessary to allow a comparison between the
comparator system and the screen system. Indeed, the mastery
of the determination of uncertainty is less obvious, especially
when calculations of characteristics are concerned (evaluation
of the propagation of errors) [13].
~114-