more 
e. 
be 
ght 
a more 
emera in 
lore recent 
! and 
involved 
mount in a 
iximum 
i with a 
• plate was 
and Stroud 
•ated through 
This 
, The most 
;he 
ibles higher 
jl lei 
*y over large 
pted Wild RC8 
craft 
ita consists 
ecording to a 
i to remove 
i and 
;a using 
¡rsion of 
il frame store 
;nts are now 
, The 
Rations was 
)P 3000 image 
sges, were 
system. One 
n'que is that 
of de-striping. This was performed by running a 
rectangular box filter across the image. The 
advantage of such a process is to produce a more 
consistent image. However, since the filter cannot 
discriminate between small scene variations and 
systematic and random noise, both tend to be removed 
or reduced in amplitude. 
For applications where there is an interest in the 
detection of subtle linear features, (for example the 
identification of subterranean field drains), it has 
been found useful to perform a variable filter edge 
enhancement, the filter size being dependant on the 
particular feature of interest. The most commonly 
used filters range from 3 X 3 to 15 X 15 pixels in 
dimension. Once the feature of interest has been 
enhanced by this process it has also been found 
useful to apply a median filter; this resamples the 
data in a 3 X 3 box. By using this process isolated 
pixels caused by the edge enhancement procedure tend 
to be removed. 
A further processing operation is required if the 
project requires quantitative absolute temperature 
values. In order to obtain such information 
calibration of the imagery needs to be carried out 
using ground temperature measurements obtained at the 
time of the aircraft overpass. Correlation of these 
ground measurements with the unstretched digital 
numbers can be used to establish the limits for a 
quantitative density slice representing discrete 
temperature levels. The application of this technique 
to heat loss monitoring is presented in section 6. 
6 APPLICATIONS 
As mentioned previously several flights of the Barr 
and Stroud IR18 TVFS were performed during August 
1984 as part of a preliminary research assessment of 
the instrument. As a consequence of early morning 
mist these flights took place from late morning until 
late afternoon rather than, as planned, during the 
post-sunset/ pre-dawn period. The imagery obtained 
was, nevertheless, useful since it enabled the 
potential of the imagery to be assessed together with 
methods of processing the imagery (Dele, 1985) 
More recently both the Barr end Stroud and RPC 
TVFS systems have been flown, for a variety of 
projects by ERSAC Ltd. The following material 
reports some of the most significant applications to 
date. 
6.1 Heat Loss Monitoring. 
Several projects have been carried out using the Barr 
and Stroud IR18 to assess heat loss from industrial 
and residential buildings. As mentioned above, in 
order to provide quantitative estimates of localised 
heat loss from TVFS imagery it is necessary to 
establish calibration data to relate the grey scale 
variations on the image to emitted radiance and 
eventually to estimates of heat emission. In addition 
to deriving emitted temperature it is also necessary 
to evaluate the variations in emissivity over the 
scene. 
Ground temperature measurements can be obtained 
using either conventional in-situ temperature probes 
(e.g. thermistors or thermometers) or alternatively 
by using ground based radiometric measurements from 
instruments such as the AGA Thermovision. As 
mentioned earlier the AGA Thermovision is a low 
spatial resolution system; however, it is possible, 
by using a blackbody reference, to obtain ground 
temperature measurements. By using the Stefan- 
Boltzmann law the heat loss for a particular 
temperature value can be derived and a suitable 
density slice for the IR18 imagery obtained. The 
results of one particular set of date are shown in 
Table 4. 
If the computed heat losses (Q) from Table 4 are 
regressed on the mean digital number indicating 
emitted radiance (R) from the IR18, then a regression 
coeffiecient (r) of 0.924 is obtained. The 
Table 4. Calibration of Heat Loss from Barr and 
Stroud IR18 using data from the AGA Thermovision 
Colour 
Coding 
Mean 
Radiance(R) 
AGA 
Temperature 
(*c) 
Mean Computed 
Heat Loss (Q) 
(Watts) 
Blue 
44 
2.5 
11 
Cyan 
74 
4.0 
17 
Magenta 
80 
4.7 
21 
Grey 
95 
6.3 
28 
Black 
110 
6.5 
29 
Red 
131 
7.6 
33 
White 
155 
13.8 
63 
coefficient of determination (r 2 ) is 0.854, 
indicating that 85% of the computed heat losses from 
the calibration results above can be explained by the 
corresponding radiance values (R) as recorded and 
digitised from the IR18 sensor data. The resulting 
heat loss regression equation for the IR18, for the 
8°c gain setting was of the form; 
Q = 0.4185R - 12.537 ....(1) 
with a standard error of ±6 watts. 
6.2 Thermal Mapping of Roads 
A novel environmental engineering application of the 
IR18 TVFS system has involved thermal mapping of road 
surfaces in winter to identify accident black spots. 
The primary objective of this night-time aerial 
survey work was to assist road engineers in the 
siting of ground sensors to monitor ice conditions 
around high accident risk sections of road. It is 
also hoped that the data will enable improvements and 
ecomomies in road gritting operations to be achieved. 
The survey was carried out in Scotland in February 
1986. Ground temperature measurements were obtained 
and used to determine calibration data for the IR18. 
By using this data in conjunction with the GEMS image 
processing system it was possible to digitise and 
calibrate a selected number of frames of data. Using 
a digital to analogue converter it was then possible 
to create from these selected scenes, a density 
sliced video image of the original analogue tape. 
This feature has been shown to be of considerable 
benefit to non specialists viewing the imagery. 
6.3 Geotechnical Site Investigations 
An assessment of the potential of the IR18 to detect 
solution features in chalk will be carried out during 
the summer of 1986. Solution features are a serious 
geotechnical hazard and generally provide unreliable 
bearing capacity for foundations. A review of the 
significance of solution features end the use of 
remote sensing techniques for their detection is 
provided in Kennie and Edmonds (1986). The relative 
performance of TVFS and thermal infrared linescanning 
techniques will also be investigated during this 
exercise. 
6.4 Drainage, Sewer Collapse and Utilities Surveys 
A recent application of the IR18 has involved the 
detection of subsurface utilities and problems 
associated with such features. In this case the TVFS 
was mounted on a boom attached to a car and was used 
in conjunction with a ground impulse radar system 
(x = 6 to 330 centimetres).