roc. 
  
  
Cruz, Santiago 
A PROGRAM FOR AUTOMATIC INNER ORIENTATION OF DIGITIZED NON METRIC 
IMAGES (35 AND 70 mm) 
Santiago CRUZ, Javier CARDENAL', Jorge DELGADO’ 
"University of Jaén, Spain 
Departamento de Ingeniería Cartográfica, Geodésica y Fotogrametría 
jcardena@ujaen.es , jdelgado@ujaen.es 
Commission V 
KEY WORDS: Photogrammetry, Software, Semi-automation, Inner orientation, Image processing, Non-Metric 
cameras 
ABSTRACT 
A program for the automatic inner orientation of non-metric digitized images taken with 35 and 70 mm cameras has 
been implemented under I.D.L. 5.0 (Interactive Data Language, release 5.0 from Research System Inc.). The program is 
based in the detection of the four edges that define the frame format. The approach to detect the edges uses Prewitt 
filters that can detect horizontal (both lower and upper) and vertical (both right and left) lines. Once edges have been 
detected, they are measured by means of transverse profiles throughout the detected edges. A regression fit line is 
computed each edge. However it is necessary blunder detection and elimination of anomalous data or noise. Then lines 
are computed again without blunders and the frame corners are calculated by intersection of lines. The photo-coordinate 
system is referenced to the center of the format (the indicated principal point defined by intersection of diagonals) and 
the X-axis is forced to be parallel to the lower format edge. Moreover there is a possibility for coordinate 
transformations (4-parameters, 6-parameters, two-dimensional projective or 8-parameters bilinear) if fixed reference 
values exist for those corners, so some film or scanner errors can be corrected. The program runs with digitized images 
in grey scale or RGB TIFF files. Images have to be scanned without any cropping because the whole format is 
necessary for the inner orientation (edges must be clearly visible and contrasted). Non-Metric cameras and desktop 
scanners, equipped with transparency trays, combined with this method can be an effective option for low cost 
photogrammetric applications. 
1 INTRODUCTION 
Since last three decades, the development of analytical photogrammetry has allowed the extensive use of non-metric 
cameras for image acquisition. Several mathematical models permit the correction of different error sources affecting 
non-metric imagery and because the low cost of these equipment there has been a wide expansion of non-topographic 
photogrammetry (Karara, 1989). Also, at present these equipment (small format non-metric cameras) are being used 
from light aerial platforms (light and microlight aircraft's or helicopters) for fast and low cost mapping applications, 
environmental and natural resources purposes, urban planning etc. (Warner, et al., 1996). But important drawbacks 
persist in the use of such cameras. These drawbacks are mainly related to the unknown and instability of the inner 
parameters, film unflatness (affecting more to 70 mm than to 35 mm film) and the lack of fiducial marks. 
Researchers have developed several approaches to avoid or minimize those problems. The unknown and instability of 
inner parameters can be solved by means of “in situ” calibration (this approach needs a dense network of surveyed 
spatial control points) or by a more effective (and complex) solution like the camera selfcalibration. Film unflatness can 
be minimized by means of additional parameters in the selfcalibration model or by installing (not always recommended) 
a réseau glass plate (see Karara, 1989 and Fryer, 1992, for a complete revision of these methods). 
Moreover, different methods can overcome the lack of fiducials. The use of the Direct Linear Transformation (DLT, 
Abdel Aziz and Karara, 1971; in Karara, 1989) is an effective alternative when requirements on accuracy are not 
essential or when the availability of a dense and adequate control point network is not a problem. Also, Faig, et al. 
(1992) propose the use of “pseudo” fiducial marks, such as slits or holes made in the focal frame at the camera back. 
Finally, another solution for the film inner orientation without fiducials is the measurement of the frame corners. This 
method can be applied even with some commercial systems like those of ADAM technology (Elfick, 1986). This last 
approach has been selected in this paper for the automatic inner orientation of non-metric imagery. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000. 149