ul 2004 
om high 
ons. [n: 
M). 
otential 
Remote 
sing of 
cations. 
:t Guide. 
vnloads/ 
ssed 27 
ation in 
gramm. 
f high- 
1omials. 
zery for 
e Sens., 
2008 
1e-meter 
to Sea 
on from 
115. 
Product 
rs pdfs/ 
; of the 
cessing. 
>ometric 
m. Eng. 
d points 
gramm. 
ACCURACY ANALYSIS OF DIGITAL ORTHOPHOTOS FROM VERY HIGH 
RESOLUTION IMAGERY 
Ricardo Passini * , Karsten Jacobsen ** 
* BAE SYSTEMS ADR, Mount Laurel, NJ, USA 
** University of Hannover, Germany 
*rpassini(Z)adrinc.com 
**jacobsen@ipi.uni-hannover.de 
WG IV/7 
KEY WORDS: High resolution, Satellite, Orientation, Matching, DEM/DTM, Orthoimage 
ABSTRACT: 
The generation of digital Orthophotos now days can be done with aerial, but also very high resolution space imagery. For a real 
competition the geometric and radiometric quality has to be on the same level. Different QuickBird images have been used for the 
generation of orthophotos with Im pixel size. The whole procedure from the orientation up to the final step has been analyzed in 
detail. Rational polynomial coefficients (RPC) and bundle orientation using orbit information was used for the handling of 
QuickBird Basic and Standard Imagery. The special problems of individual and combined scenes are analyzed. 
QuickBird Images covering quite different areas were used. The effect of varying control point distributions on the accuracy, 
determined with independent check points was studied. Different sources of ground control like digital orthophoto quads, existing 
information from airborne photo flights and GPS-control points have been used. The required height information came from 
different digital elevation models (DEM) and ground survey. Satisfying results have been achieved. 
1. INTRODUCTION 
Accepting the rule of thump for which in topographic maps a 
pixel size of 0.05 to 0.1 mm in the map scale is required, with a 
Ground Sample Distance (GSD) or a pixel size in the terrain of 
61 cm a map at a scale up to 1:6,000 can be designed. Also 
accepting the fact that in an ortho-image there should be a 
minimum of 8 pixels/mm (otherwise the pixilation becomes 
visible), then based on a GSD of 60 cm the ortho-image design 
scale can be as big as 1:4,800. 61cm is the nominal GSD of the 
QuickBird B/W Panchromatic band for a nadir view, hence 
when working with this satellite imagery, maps up to a design 
scale of 1:6,000 and orthophotos of 1:4,800 are possible. In 
addition to the aspect of information contents, also the 
geometric potential is important. This is depending upon the 
precise identification of objects in the images and the image 
geometry itself along with a sufficient mathematical model. 
In the present investigation different available mathematical 
models for the ortho-image generation using QuickBird 
imagery have been studied. Different sources of Ground 
Control Point (GCP) information, different number and 
distribution of GCPs, different number and arrangement of 
images and different environmental scenarios (i.e., dry urban 
desert, humid farming areas) have been included in the study. 
2. SENSOR INFORMATION 
Basically the QuickBird Images are of line scanner type. As 
such the image geometry is central perspective in the line 
direction. In this sense the exterior orientation parameters of 
each line are different, but the relationship of the exterior 
orientation to the satellite orbit is only changing slightly. Hence 
for the classical CCD-line cameras, the attitudes are not 
changing in relation to the satellite orbit. The Earth is spinning 
in this system. The projection centres are located in the satellite 
orbit — this can be expressed as a function of the image 
components in the orbit direction. 
The new generation of sensors has the flexibility of changing 
view direction while acquiring the image. In this sense the 
sensors can change continuously the view direction in such a 
way that their image lines are located parallel to local or 
national East — West map projection grid direction. This is a 
continuous on the orbit change of the yaw and roll movements 
to reach the scene border line with a fixed east value. This is 
shown in Figure 1. 
  
  
Figure 1: imaging 
geometry of the very 
high resolution satellite 
systems with flexible 
view direction 
  
  
  
  
  
695