372
Real -Time Correction Of Geometric Distorsions in Airborne Remote
Sensing Systems
H. Hirsch, J. Nopirakowski, M. Scheele, Th. Terzibaschian
Institut for Space Researche, Berlin
Rudower Chaussee 5
Berlin
1199
Dr. F. Plischke
INTERFLUG, Betrieb Fernerkundung, Industrie und Forschungsflug
Berlin-Schoenefeld
1189
Images, produced by an airborne electro-optical line scanner,
show geometrical distorsions. They have to be corrected in a
strong way line by line with respect to the airplane motion.
These disturbances can lead to displacements between neighbouring
lines of the recordered images. They are eompareable with the
image-pixel size. The topic of this report is to present a
possibility to correct it by using attitude-measurements,
especially measurments of the angular motion around the three
body axes or measurments of the three tilt angles yaw, pitch and
roll. These angular motions are connected with typical image
distorsions. The roll-angle produces a lateral shift of image
lines and decreases so the usable swath-width. The yaw motion
rotates the image lines and the pitch-angle produces varying
distances between neighbouring image lines. The rather complex
angular airplane motion produces ergo rather complex geometric
distorsions./1/ The authors' aim is the development of inteligent
real-time onboard control algorithms for elctro-optical line
scanners. It shall be demonstrated studying the pitch-angle
related distorsions. One way to correct the pitch-effects is a
control of the line's sample time. Because of the velocity v,
the altitude h and the pitch motion the projection of a sensor
pixel onto the earth moves during a discrete time interval AT
over a distance Ax :
Supposing quadratic pixels and realistic pitch-angles of about
some degrees, the right sample period AT 1 , the time between two
following read outs of the sensor line, can be derived using
formula 1. It is possible to aproxímate
Here isAT^the normal sample period under ideal conditions, i.e.
M )
дх s v дт + h• tonfi-AT)
tan (*S• ДТ)
Thats why formula 2 is :
