scanner imagery, and scanning laser altimetry data As a part 
of the laser altimetry suit precise GPS positions and INS atti- 
tude have also been recorded. We plan to make this data set 
available as a part of an ongoing effort of WG [11/5 to estab- 
lish standard data sets against which different data analysis 
techniques can be tested Csatho et al (1998) describe the 
data set in more detail 
Digital elevation data were acquired by the Airborne Topo- 
graphic Mapper (ATM) laser system The ATM is a con 
ical scanning laser altimeter developed by NASA for pre 
cise measurement of surface elevation changes in polar ice 
sheets, ocean beaches and drainage systems (Krabill et al 
1995) The multispectral data were collected by the Daedalus 
AADS-1260 airborne multispectral scanner from the National 
Geodetic Survey (NGS). The AADS-1260 is a multispectral 
line scanner with eleven spectral bands in the visible, near in 
frared and thermal infrared. The system is also called airborne 
MSS since it has been created for airborne simulation of the 
Landsat MSS satellite system. Aerial photography was ac- 
quired with an RC20 camera, also from NGS. The laser scan- 
ner and the multispectral scanner were mounted on NASA's 
P-3B aircraft. The aerial camera was operated independently 
by NGS, but on the same day 
First, the raw data were somewhat preprocessed. For ex- 
ample, the GPS, INS, and laser range data were converted 
into surface elevations of laser footprints The laser scan- 
ning system covers a 200-300 meter wide swath with a set 
of overlapping spirals To find an interpolation methods that 
gives satisfactory results in case of a urban scene with lots of 
discontinuity is not a trivial task. We used a thin plate spline 
method. The multispectral data should be corrected geomet: 
rically and radiometrically. A simple scheme based on Chavez 
(1989) was used to correct for the atmospheric effects 
5 Examples for using multispectral data in object 
recognition 
In this section we present examples that illustrate the po 
tential of using multispectral methods for object recognition 
The aerial photographs were used as ground truth during the 
interpretation Multi- (and hyperspectral) systems are cap- 
turing images in a number of spectral bands in the visible 
and infrared region In this part of the EM spectra, except 
for the thermal infrared, the dominant energy source is the 
solar radiation, and features in the images are mostly related 
to changes in surface reflectance, or in the orientation of the 
surface elements or in both Owing to the complex rela 
tionship between the spectral curves of the surface materials 
(e g., vegetation, concrete, water) objects may look quite dif 
ferent in images acquired with different wavelengths As an 
example compare the visible and NIR images of the residential 
area in Figure 1 and 2 The roof of building 'A' has enough 
contrast to be distinguished in both images The light roof 
of building 'B' clearly stands out in the visible image, but it 
completely disappears in the background in the NIR image 
The dark roof of building 'C' represents the opposite case. it 
can only be recognized in the NIR image 
Images in near infrared It is well known, that the con 
trast between natural surfaces such as bare soil, grass, water 
and urban structures, eg asphalt concrete, etc, is gen 
erally higher in infrared than in visible (eg , Jensen, 1983) 
Non-turbid, deep. clear water almost completely absorbs the 
infrared energy resulting in a very low reflectance (Figure 2 
itm uem 
Vtt aite, i) (ORA. LEE ttm Lm, 
FRIIS prety a 
  
Figure 1: First Principle Component of six visible bands over 
a residental area in Ocean City, MD (Daedalus MSS, Bands 
2-8. 0 42-0.69 yum, pixel size is appr. 0.8 m) Marked objects 
are buildings (A, B, C), empty lot (D). road (1), channel 
(2). and trees (3) 
re 
3 
À 
7 
ER Y 
1 d Lid gei 
^ E 
  
Figure 2 Near infrared image (0 8-0 89 yim) 
2') Live green vegetation scatters most part of the solar 
radiation For example, the bright area around building ‘A 
is most likely associated with vigorously growing lawn (Fig- 
ure 2) There is almost no energy reflected back from areas 
in deep shadow This causes very low gray values and a con- 
siderable loss of details 
338 International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 
Im. 
the 
'he 
sur 
tru 
froi 
fact 
Bec 
(e € 
can 
will 
con 
coo 
higl 
at r 
hibi 
Pav 
and 
and 
Dur 
age 
age: 
and 
driv 
In € 
have 
scer 
plet 
very 
met 
trem 
Nor 
IS NC 
spec 
face 
Veg: 
tool: 
chlo 
vege 
the | 
mati 
in th 
on tl 
Vege 
relat 
indic 
bility 
Figu 
color 
with 
segm 
exair 
Clas: 
unsu 
OUS ( 
per ( 
meth 
most 
sified 
and | 
(C2) 
house