y, Y., 
ution 
DER 
r2. 
ng in 
-610. 
in, M. 
kness 
nal of 
light 
., 38, 
AIRBORNE REMOTE SENSING EXPERIENCES 
WITH DIRECT PLATFORM ORIENTATION 
Charles K. Toth 
Center for Mapping 
The Ohio State University 
1216 Kinnear Road, Columbus, OH 43212-1154 
E-mail: toth@cfm.ohio-state.edu 
Commission VII, Working Group 1 
KEY WORDS: Sensor Integration, Digital Imaging Sensors, GPS, INS, Airborne Mapping 
ABSTRACT 
Unprecedented technological advancements in generic computer hardware and sensor technology have created numerous 
new mapping methods and redefined the entire spatial data extraction methodology in the last few years. These changes 
have dramatically impacted the remote sensing and mapping industry, including almost every level of the spatial data 
acquisition, processing, and dissemination process. 
Airborne remote sensing has long suffered from the lack of adequate and affordable platform orientation. Although the 
accuracy requirements in this field have always been somewhat relaxed compared to topographic mapping, the need for 
more precise georeferencing has started to grow recently. Increasing use of integrated Global Positioning Systems (GPS) 
and Inertial Navigation Systems (INS) as a core aerial platform orientation component will provide a powerful new tool 
for georeferencing modern airborne remote sensing and mapping systems. | 
1. INTRODUCTION 
Modern airborne multi- and hyperspectral systems are 
based on electronic imaging technology. Typically line 
or area Charge-Coupled Devices (CCD) form the heart 
of the sensory system complemented by analog or 
mostly digital data acquisition components. There has 
been a variety of image formation methods such as 
pushbroom, scanning, three-line, and frame camera 
geometry models. Hyperspectral systems usually scan 
across a line on the ground, generating a large number 
of pixels in each scan line while aircraft motion 
provides the second direction to form an image. The 
optical projection system splits the incoming ray into 
many narrow spectral components for each pixel, 
adding a third dimension to the acquired data. In 
contrast, state-of-the-art multispectral systems are based 
largely on a frame camera model and come with a less 
sophisticated optical projection system, splitting the 
incoming beam into a relatively few wide spectral 
channels. 
Multi- and hyperspectral imaging systems provide a 
new extension to airborne remote sensing and mapping 
technologies by providing an additional data dimension 
to the spectral signature. Due to the unparalleled 
advancements of generic computer hardware and sensor 
technologies, there have been a large number of 
airborne multi- and hyperspectral systems developed 
since the early 1990s. An excellent review of 19 
airborne hyperspectral systems is given in (Birk and 
McCord, 1994). In fact, the rapid emergence of these 
systems has far outpaced the rate of how applications 
are being found for the data, coming in large volume 
from these sensors. The relatively high price of the early 
airborne hyperspectral systems has also been a major 
obstacle in the commercialization of the multi- and 
hyperspectral imaging technologies. At present, with the 
significantly improved expertise and with dramatically 
falling hardware prices, combined with increasing 
performance, applications are about to enter map 
production. An already commercially successful system 
in the multispectral market is the ADAR system from 
Positive Systems. AISA is similar hyperspectral system 
from Finland, see (Mákisara, 1993). 
An analysis of the system structure of the multi- and 
hyperspectral systems can easily reveal that the 
georeferencing task of the image data has played a 
somewhat secondary role at the beginning. This was due 
to two facts: (1) hyperspectral imaging itself was quite 
new, and (2) high-precision platform orientation was in 
its infancy (GPS constellation was not complete and 
high-precision INS systems were unaffordable in the 
early 1990s). Over time, however, the demand has 
started to grow toward more accurate georeferencing of 
captured data. GPS has been a key component in this 
process by providing precise geographic location data 
(latitude, longitude, and altitude) of the aircraft at the 
time images are acquired. Although this solution was a 
major step ahead, it had serious limitations. First, GPS 
positions were available only at distant time epochs 
(typically once every second). Especially for 
hyperspectral systems, where continuous recording is 
mandatory, this presented an uneven accuracy pattern of 
location data. Second and more importantly, a single 
GPS itself could not provide the aircraft attitude data 
crucial in high-precision georeferencing. Obviously, the 
availability of location and attitude data can 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 47 
f 
E 
1 
n 
x 
Eu 
E 
p 
M 
"tH 
E 
nut 
A 
s 
P 
ul 
mj 
É 
E 
A 
I^ 
^ 
i 
E 
M. 
x 
* 
iu 
Hi 
x] 
|^ 
HJ 
3 
I 
ie 
i 
| 
RR 
2e 
crc