stanbul 2004 
formation of 
ment may be 
e time series 
segmentation 
mage (Baatz 
] and Multi- 
c. of the 2nd 
of Remote 
vw.definiens- 
(accessed 27 
iate optimal 
/ Journal of 
Rudant, J.P., 
nultitemporal 
Rudant J. P., 
images. Proc. 
D., Taconet, 
Land cover 
mages and 
1 agricultural 
nsing, 21(3), 
d Meloni, M., 
ion technique 
mages. IEEE 
, 41(11), pp. 
$ P. 2000. 
sh landcover 
'e Sensing of 
. and Ohtomo, 
s Analysis by 
ress, Japan. 
DESIGNING AND DEVELOPING A FULLY AUTOMATIC INTERIOR ORIENTATION 
METHOD IN A DIGITAL PHOTOGRAMMETRIC WORKSTATION 
M. Ravanbakhsh 
Surveying college, National Cartographic Center (NCC), Tehran, Iran, P.O.Box:13185-1684 
Email address: Ravanb@ncc.neda.net.ir 
Commission IVIV 
KEY WORDS: Orientation, Accuracy, Digital, Transformation, Aerial, Pixel 
ABSTRACT: 
This paper is concerned with a comparative study and implementation of different image correlation techniques for Automatic Inner 
Orientation in aerial images. 
The implemented image correlation techniques are: 
Cross Correlation Function (CCF), Binary Cross Correlation Function (BCCF) and Least Square Matching (LSM). The first two 
approaches are used to determine the approximate position of the fiducial mark centres, which is then followed by a quadratic 
surface fitting for precise fiducial centre determination. Wiener and constrained least square filter were used as pre-processing 
techniques to improve the accuracy of algorithms. 
All three algorithms are applied to the digital images with different resolution (namely: 15, 30 and 60 micrometers pixel size). 
The test results show that the optimum solution can be achieved by a combination of CCF for approximate positioning followed by 
LSM method, i.e. LSM approach shows better performance as far as pointing precision for fiducial mark(cross type)centres have 
been 4.2 and 4.7 micrometers respectively when the optimum solution (i.e. CCF plus LSM)is used. Other aspects of fully automatic 
Digital Inner Orientation with respect to image frame direction, and determination of whether or not a mirror image is scanned are 
also investigated. The designed algorithms have the capability to achieve the accuracy mentioned above in any frame from different 
aerial cameras. 
1. INTRODUCTION the matrix including grey values but photo coordinate system is 
defined by fiducial marks positions. If the relationship between 
Although Several decades have passed since scientists in the scanner and camera coordinate system remain constant, inner 
photogrammetric community started research work related to orientation will be eliminated from digital photogrammetric 
automation of different processes in photogrammetry and stages. We have not had the chance of using such digital 
remote sensing, with the dramatically fast progress of computer cameras to prepare high resolution images yet. 
technology, great developments have occurred in recent years. We would be able to design an algorithm to achieve the goal of 
With precise scanners and high memory fast performing performing interior orientation automatically if it is possible to 
computer systems, the possibility of automation in Digital have a good knowledge of location, shape, illumination 
Photogrammetric Workstations (DPW) has increased. New distribution and sizes of fiducial marks in digital scanned aerial 
techniques such as machine vision and digital image processing photos. With respect to fully automation procedure concept, the 
and the trend of commercial photogrammetric companies in algorithm should be able to include features listed below: 
using DPW made the research grow quicker. 1- using images from various cameras 
Inner orientation is a prerequisite for any project including 3D 2-using colorful and black & white images 
computation as it is a complicated and time-consuming process 3-using fiducial marks with different of shapes 
which making it automatic helps us to open a broad range of  4-using images with different pixel size 
applications. Among orientations (inner, relative and absolute) 5-using positive or negative images 
inner orientation has a special importance as any inaccuracy 6-using mirror images 
will affect next stages in photogrammetric processes. 7-using rotated images 
In 1995, Phodis Company developed a method of fully 8-using images scanned in different scanners 
automatic inner orientation with overall accuracy of 0.2 pixel 
and pointing accuracy of any individual fiducial mark of 0.1 3. DESIGNING AND IMPLEMENTING 
pixel. 
With respect to the kind of input data, several methods can be 
2. INNER ORIENTATION AND AUTOMATION taken into account, however, two distinct algorithms performing 
FEATURES two major successive stages of localization and precise 
measurement differently and other minor six stages commonly. 
In inner orientation process, we establish a geometric The six common stages are: 
relationship between photo coordinate system and instrument |-extracting image patches 
coordinate system. In metric imaging system, photo coordinate 2-resampling the template 
System is defined by fidcual marks but in field of digital 3- Image pyramid derivation 
photogrammetry instrument coordinate system is replaced by 4- Detection of the orientation of the image 
pixel coordinate system. Image coordinate system is defined by 5-positive-negative recognition 
un 
Ax 
2