is obvious
very end of
a quality is
nt or it's
; case there
ollected for
some other
rivate firms
ovided, not
for quality
other hand,
imizing its
so through
ess.
99. Project
d Mapping
ometrischen
)issertation.
ft Nr. 539
senschaften,
nt. Juran's
cGraw-Hill,
e Japanese
4ssurance —
— Quality
— Quality
n Approach
ion, chapter
nited States
agement in
ications in
| Procedure
s, ITC, The
SPATIAL AND RADIOMETRIC QUALITY OF THE MOSAIC OF IMAGES ACQUIRED BY AIRBORNE DIGITAL
VNIR MATRIX CAMERA AND TIR LINE SCANNER
A. Krtalic*, T. Fiedler**
*University of Zagreb, Faculty of Geodesy, 10000 Zagreb, andrija.krtalic@zg.hinet.hr
**University of Zagreb, Faculty of Geodesy, 10000 Zagreb, teodor.fiedler@zg.htnet.hr
KEY WORDS: mosaic, geocoding, digital parallel scan camera, digital matrix camera, geometric deviation, radiometry deviation
ABSTRACT:
The analogue aerial photo cameras are using the panchromatic, color visible, color infrared film and dominate in wide area
photogrammetric data acquisition but also digital cameras became available on the market. Digital airborne sensors offer new
technical and opperational opportunities although require new approaches, knowledge, education and training. While very suitable
and efficient in use the digital airborne sensors bring new problems that should be solved for best quality of their products. In this
paper we consider the quality of spatial and radiometric features of the mosaic of images acquired by two digital airborne sensors
that have different modes of image formation and work at different wavelenghts. The first sensor is the digital matrix (staring)
camera, for three visible channels (V: 0.4-0.5 pm, 0.5-0.6 um, 0.6-0.7 um), for near infrared channel (NIR: 0.7-1.0 um) and the
second sensor is the longwave thermal infrared (TIR: 8-14 um) parallel scan camera. Although for TIR wavelengths exist matrix
cameras they have rather limited resolution (320x240 pixels) and the parallel scan cameras are in intensive use due to better
resolution. The both sensors are in intensive operational use in Croatia, and this was the motif for this analysis. We analyse spatial
and radiometric quality of the mosaic of high resolution VNIR images (1392x1040 pixels) and of TIR images (600x400 pixels). If
compared to VNIR images, TIR images have spatial distorsions due to parallel scanning mode. For the case study was selected the
scene that has small number of objects that could be used for registration of TIR images onto VNIR images. The work is part of the
research conducted in the scientific project ARC funded by European Commission.
1 INTRODUCTION 2 GEOMETRIC COMPARISON OF IMAGES
Operative system which consist of two sensors, two digital
cameras, DuncanTech MS3100 and Thermovision THV 1000,
2.1 Selection of a pair of images and equalizing of scale
are in intesive operational use in Croatia for remote sensing of
minefields. The first sensor is the digital matrix camera, for
acquisition in three visible channels (V: 0.4-0.5 um, 0.5-0,6 um,
0.6-0.7 um) and near infrared (NIR: 0.7-1.0 um) and the second
sensor is parallel scan camera which collects data in the
longwave thermal infrared (TIR: 8-14 um) area. These two
cameras are different in many ways, they collect data in
different wavelengths and they have different principle of
collecting data. DuncanTech MS3100 is the digital matrix
camera (sensor resolution 1392x1040 pixels) and it produces
images in central projection. Thermovision THV 1000 collects
the data by scaning 5 parallel lines at the time in 80 rows with
resolution of sensor 5x400 pixels. The reason for this particular
camera to be used instead of matrix camera (available in the
market) is the sensor resolution. The matix TIR cameras have
resolution of 320x240 pixels and for that reason parallel scan
camera is in use. The flights for collecting the images for the
purpose of humanitarian demining in Croatia took place at the
height of 130 m and higher above the ground. Because of small
flight height, the small surface of each image and a large
number of images, the orientation in space has become more
difficult. This problem can be solved by mosaicing the selected
images of the an area and geocoding of the whole mosaic.
However, these proceses generate new problems, as additional
geometric and radiometric deformation of images made by
interpolation in mosaicing and geocoding. Therefore, for the
practicle use of mosaic, it is necessary to know what can we
expect from it with respect to geometric improvement and
radiometric distortions. This is important especially for TIR
mosaic because of considerable geometric deformations of
original (input) images. For that purpose the comparison of
geometric and radiometric relationships between original
images and geocoding mosaics was done.
For geometric comparison there was a pair of images selected
(one VNIR and on TIR) with the biggest owerlap between
images and the biggest quantity of details which can be detected
on both images. The end points of both images were defined by
their sensor resolutins and images were interactivly positioned
between those points. Thus, a common unit of measurement (for
both images) is a pixel. On both images there were 8 identical
details identified so that certain details on VNIR image can be
joined with two parallel horizontal and two parallel vertical
lines (lines that are parallel with the axes of coordinate system).
2 pi
Figure 1. Delineated identical points on the pair of images,
VNIR (left) and TIR (right), with 6096 overlap
The lines between the same points on TIR image aren't
horizontal and vertical because of parallel scaning and moving
of platform (helicopter), they are inclined. Diferent resolutions
of sensors for comparison those two images and true value of
deformations between them call for equalizing of the scale (the
most approximate scale). It is achieved by projecting the
inclined lines of TIR image on to horizontal and vertical lines
(parallel with coordinate axes) on VNIR image. After that, the