er reliable 
scher M. 
Monitoring 
intelligent 
Shanghai, 
iption.html 
1 airborne 
eir fusion. 
roceedings 
echnology 
atic early 
and Fire 
ntegration, 
Wichmann 
28. April 
ne remote 
european 
rence On 
CD-ROM, 
M., Dalaff 
echnology 
on Stereo 
ce. Direct 
canner. — 
ace 2001. 
and of 
Xf Optical 
OS S.A. 
A. New 
eof the 
FhG IPK, 
Wireless, 
' [nstitute 
ogy and 
mtechnik 
anbul 2004 
  
  
  
PROCEDURES FOR RADIOMETRIC QUALITY CONTROL OF SCANNED CIR 
IMAGES 
L. Markelin 
* E. Honkavaara 
Dept. of Remote Sensing and Photogrammetry, Finnish Geodetic Institute, Geodeetinrinne 2, P.O.Box 15 Fin-02431 
Masala Finland — lauri.markelin @fgi.fi, eija.honkavaara@fgi.fi 
KEY WORDS: Radiometric, Quality, Infrared, Scanner, Aerial, Image 
ABSTRACT: 
In state-of-the-art orthophoto production aerial films are routinely scanned to a digital format using roll-film scanners. Scanning 
parameters for a large number of images are typically set subjectively based on only a few images and quality check is done 
visually after scanning. For further use, orthophotos are often radiometrically enhanced, but the criteria for these enhancements are 
mostly subjective. The objectives of this study were to develop tools for the radiometric quality control (QC) of the scanning 
process and to investigate the tone tuning process of colour infrared (CIR) images. The central method of the radiometric QC is a 
100% histogram control. The histograms of all scanned images are calculated during the scanning process and compared to 
existing tolerance values. These calculations also aid in the scanning parameter selection. In this study, the tolerance values were 
determined based on histograms of a total of 2818 images taken using four types of film. The conclusion was that the main 
statistics of image histograms were efficiency, 99%-efficiency and saturation. Experience has shown, however, that also a visual 
check of images is essential in addition to the automatic histogram control. Based on extensive investigations of radiometrically 
enhanced orthophotos, a tone-model-image for the tone tuning process was created. 
1. INTRODUCTION 
The Ministry of Agriculture and Forestry (MAF) of Finland 
maintains Finnish Land Parcel Identification System (FLPIS), 
due to the European Union’s (EU) demand on controlling the 
agricultural subsidies. The FLPIS is a Geographic Information 
System containing location information of all parcels and 
farmsteads of farmers that have applied for area-based 
subsidies. A central component of the FLPIS is a countrywide 
orthophoto database. 
The first FLPIS orthophoto mission was executed in Finland in 
1996-1997. Three contractors with different systems produced 
the orthophotos from existing 1:60 000-scale panchromatic 
images. Finnish Geodetic Institute (FGI) functioned as the 
quality control (QC) consultant. The quality and production of 
those orthophotos has been thoroughly discussed by 
Honkavaara er al. (1999). The five-year update process of the 
first orthophoto series began in 2002. Several contractors are 
involved in the production. The orthophotos are produced from 
1:31 000-scale colour infrared images (CIR) with 0.5 m pixel 
size. CIR images were selected to enable the use of the same 
material in forestry applications. MAF decided to apply a 
comprehensive QC-strategy in the process; FGI created the 
quality system (Honkavaara 2003) on the basis of the European 
Commission's (2004) recommendations. The FLPIS orthophoto 
production consists of two stages: digital image production and 
orthophoto production. The QC is divided to an internal control 
of the contractor and to an external control organized by the 
customer. An external quality consultant company performs the 
external QC. 
  
* Corresponding author 
Radiometric quality affects significantly the interpretability of 
images. Careless treatment may lead to a severe loss of 
information content. In the FLPIS orthophoto production, after 
the imagery flight and the film development, the radiometry is 
treated in the scanning process and in the mosaicking and tone 
tuning processes 
An efficient tool for the radiometric QC of digital images is a 
100%-histogram control. The idea of the histogram based QC 
is to calculate the histograms and histogram statistics of all 
scanned images and to compare the statistics to the determined 
tolerance values. The best efficiency of the method is obtained, 
if it is executed immediately after the scanning process. This 
method has not been used yet in the FLPIS QC-system, 
because the tolerance values have been missing. European 
Commission (2004) gives recommendations for saturation and 
contrast of luminance-histogram, but these were considered 
insufficient for the CIR-image based FLPIS process. 
Histogram control cannot completely replace the visual check 
of images. First of all, abnormal histogram properties may be 
caused by some acceptable phenomena (e.g. large waters). 
Because of this, the histogram control software should also 
collect thumbnail images. The second reason for the need of 
the visual inspection is that consecutive image enhancements 
may result in artefacts that can be noticed only visually. In the 
FLPIS process, the external QC checks a sample of images, 
selected according to ISO 2859-standard, in the acceptance 
control by interactively viewing the image with a feasible 
magnification. The QC-strategy of the FLPIS image production 
process is presented in Figure 1. The orthophoto QC has a 
similar structure (Honkavaara 2003).