International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 In. 
  
  
information (Figure 1) The reconstruction is based on the 2.5 SilverEye TM. An Integrated RFM Based Environment 3. 
identification of two conjugate points in the stereo model. Using 2 
their image coordinates it is possible to express the relation SilverEye is a software tool that has been developed based on n 
between the left and right image coordinates and the 3D the RFM scheme. It has been commercialized and marketed by st 
coordinates of the corresponding ground point. GeoTango International Corp. (www.geotango.com). It utilizes In 
the RFM as the internal geometry model for various th 
photogrammetric mapping tasks (such as ortho-rectification and th 
Fc stereo feature extraction). As high resolution satellite products, at 
such as Ikonos and QuickBird are delivered with RPCs the T: 
n photogrammetric process is made transparent to the users: they hc 
. . are no longer required to master a physical sensor model and Sk 
s / perform interior or exterior orientation. Once an image is SW 
\ 7 T loaded, the corresponding RPCs are uploaded, and users are * 
m 
able to start performing their tasks immediately. SilverEye 
provides a mono (see Figure 2a) and stereo (see Figure 2b) 
working environments. In each of these environments users are 
? t able to collect 2D data and 3D data and organize it in a GIS- 
style layer format. T 
   
  
  
  
3.1 
Figure 1. 3D reconstruction modes using the REM Hs 
i 0 
Let X, Y and Z to be the un-normalized coordinate values of i cor 
points in object space. The normalization of the ground : PC 
coordinates is computed using the following equations: ^ Th 
cor 
x ed s. ede zi. (3) fro 
X Y Z for 
' s s Fig 
where X, Y, and Z, are offset values for three ground 
coordinates, and X,, Y, and Z, are their scale values. The first- The 
order approximations are obtained by applying a Taylor Tat 
expansion of /and s towards the three input variables X, Y, Z all 
in Eq. 1. Thus, when considering the normalization parameters, effe 
the four error equations for two conjugate image points (/, sj) bot 
and (/,, s,) are derived: star 
ster 
w] [on/axiX, e/0YHY, OLIOZIZ,| p v [hb oth: 
vu Os, /OX1X, Os, OY/T, Os 10/14, AT 5, -$, The 
9.1 |arsax,x, a,sorsy, at /ezIz, rd j 
AZ rh com 
Y Os IOXTX i 085,7 0Y7Y, 6s, /02/Z .. $, 3, whe 
inte 
v="Ax ~1/ (4) Silv 
mea 
The reconstruction process begins with the estimation of the that 
ground coordinates (Xo, Yo, Zo) obtained by solving the RFM with (Fig 
only constant and first-order terms, by using one image and a E 
given elevation value, or by setting to be the offset values of the 
ground coordinates (Hu et al. 2004). Once retrieved, the 
corrections to the estimated ground coordinates are given by: 
x=[AX AY AZJ «(4 WA)  AWI 6) Figure 2b. The stereo image 3D mapping interface en 
M 
where W is the weight matrix for the image points. The weight S 
matrix may be an identity matrix when the points are measured on 3. EXPERIMENTAL RESULTS M 
M 
images of a same sensor type. However, higher weights should be 
assigned to points measured in images of higher resolution when 
doing hybrid adjustment integrating images acquired by different 
sensors and thus having different resolutions (Hu et al., 2004). 
Using the SilverEye package, various tests have been carried 
out in order to estimate its performance in two commonly used 
tasks: distance measurements and building height measurement. 
1024