For the aluminium test surface approximately 7000 3D points parameters of the exterior orientation of an image are also 
; have been acquired by the stripe projection system, cf. fig 3. determined, an image ray can be described completely in object 
pared for control leven digital images have been taken with a Gi i ined 
100 mm? and the Furthermore, € gita. Image: space. Given a second image and a corresponding image ray a 
resolution of 0.2 mm per pixel in object space. 3D point in object space can be determined. Usually, 
ecisely defined by 
: dine i ue 
action of the ded corresponding image rays are identified by analysing patches of 
image data by means of correlation or least squares matching 
  
   
production can be REV ue re EE techniques. 
vices used but as An A, i m S ri ; 
e e Dna > bee = e UM 
o assumptions will Fron, 4-7 2 i i 
p Era At nt a LH If more then two images of the same surface area are available 
de tee ee *o:ot RE : ; : ; ; ; 
: svt f RnLuensc. ug SUR GM 025: M qns. multiple- image matching may be used in order to improve the 
A : se eS LSU. PRL ea tk 
paque but isotropic Bent HED Te ny Nae JT a EEL . result 
sor Eo C MV Lc ELT ME CM eR À ; 
e the surface for 70 te UN. ER ES rA LM 
Si 2 an. tot omoia, OE m oap dE mie i nahe sn e 
' has been utilise b enitn e den Ant A NE LTR aE i inle-i ing i 
d la pt Air! EL Wiist6Ó i onc«AQE Here, the formulation of multiple-image matching is performed 
levice. dod P der a ne 2a by facet-stereo vision, which can be considered as a basic 
a of RRR es : : / : : 
IA nie moz 5 se JE concept using image data for object reconstruction, (Wrobel, 
ete an us, has Uh ssp Cea A N x inci isi i 
Su í E , 3 unm ; ER f 1987). The principles of facet-stereo vision are briefly described 
ei ttr t zs « ON E Lara . . 
| the first object, ci. fen. $t. xt B PEE. etg atti ers in the following. 
25 x 125. mm?. An PEER 2e 1 32e TE 1 
juld be a real 3D ELE ee rtr uer 4 12 you is 
Past v onu Senna nt NÉ ES MOTS The basic equation of facet stereo vision 1s 
as been used. From m poem te WN wT hl lj LA 
|. Furthermore, the Eu tL LE eu ru op rap fe 
by projection EC. gatit il Bans PY LE fe G(X,Y) z TIG 5, y)]-2 T'IG' 6, y)] 9 (1) 
; BL'LegàgnA. e Ql AS LL Er MY 
E. lm ae. to? iris TELE martial ta, RN 
at * uei r£ sv. rE » . = x s ut RUE ud h : : : : : n «c 
Sa eat fle Fin ap BIR where G is the radiometric model in object space and G*, G**,... 
wr RE Et Re, it on Le yoy io ; ; 
t vlt n Rem lot nz AIA Bik are digital images of the object. These images are transformed 
A TE RC T ASS ; ; : ; : 
BAT dmn. ti. RE res e E et a] by linear transfer functions T¢, T**, ... to the radiometric object 
ies manna? io e qu w— m S A A y UE e." 242" DC 1 1 1 
JE. RN tie ME EE " m Morus 2. model G(X,Y). The radiometric model G is calculated from an 
Puoi ii ee dci epe uoo ER, approximation of the geometric model Z(X,Y), hence 
SAEs : approximately known. 
Figure 3. Distribution of 3D points on test surface pp y 
Random errors will cause residuals between the radiometric 
model G and the original observations, the picture elements of 
the Images G', G5, .... 
On the concrete tile approximately 83000 3D points have been 
acquired and seven photogrammetric images were taken. In (fig. 
4 and fig. 5) it can be seen clearly, that due to mirror-like 
reflection there are areas, where stripe projection could not 
acquire any data. 
  
irface 
1e stripe projecting 
tische Messtechnik 
n with a consumer 
sation of free form 
two CCD-cameras. 
measured, different 
used. From this, 
ry from 0.06 to 0.6 
mm, (GOM, 2002). 
he system is due to 
sed by temperature 
jecting unequivocal 
cted patterns which Figure 4. Distribution of 3D points on concrete surface The resulting error equation is 
    
Figure 5. Intensity image of concrete surface | 
  
Yan 00 5 Y)- T [G EY) (2) 
ji Finepix S1 Pro 2.3 Explanation of the method GG 
cture elements. The 
nd was calibrated For processing radiometric information of object surfaces which is used to find the solution for geometric model Z(X,Y) 
photogrammetric approaches are to be used. On the premises inan iterative process under the condition 
that the interior orientation of a sensor is given and that the 
—107—