raw 
  
œ 
> 
o2 
mi 
Of On 
SS 
I 
B 
Un 
5 
L2 
c. 
= 
  
= 
o 
= 
La 
= 
^ 
N 
CA 
No 
LL 
Cn 
N 
N 
m 
— 
Q^ 
Os 
© 
| 
O 
CA 
_— 
Um 
| 
  
— 
+ 
A 
~~ 
= ga 
c 
= = 
e 
£5 
uan 
al © 
wi 
al 
Bl 
  
AN 
d 
“a 
| 
  
  
  
  
  
  
Comparison of Standard Deviations in overlap area of corrected images 2b and 2c 
2b band1 2c bandl 2b band2 2c band2 2b band3 2c band3 2b band4 2c band4 
Minefloor 3:7 3.5 2.9 2.7 2.4 1.9 4.6 4.5 
Regen 1 5.0 4.7 3.6 3.7 3.4 3.3 7.7 7.1 
Native 33 4.7 6.8 53 4.1 3.5 15.3 13.9 
Regen 2 4.6 4.4 5.4 4.7 4.6 4.0 7.1 6.3 
  
  
  
  
  
  
  
  
  
  
  
Area Summary (number of pixels 
Frame 2b 
Mine floor 
Regeneration 1 
Native vegetation 
ion 2 
Note that whereas the 2b and 2c means differ markedly before 
correction (Table 2), they agree closely after the correction 
procedure has been applied (Table 3). The values in the 
overlapping areas in the raw data were band-dependant and also 
varied with the nature of the materials. Differences were in the 
order of 15-30% between the bright and dark sides of the 
images. The correction produced coherence in the overlapping 
areas of better that 196 within a very low standard deviation. 
The reflectance of a range of invariant targets (compared with a 
BaSO, Lambertian standard) using an Exotech field radiometer 
(Table 1) were made during the acquisition of the DMSV 
images. The mosaicked images were converted into reflectance 
values to provide a comparable measure of the vegetation status 
and the ability to compare between the two minesites over time. 
Classification into vegetation associations was done on the 
mosaicked images with reference to field data. Fourteen 
different vegetation classes were established across the Andoom 
minesite. Statistics derived from these classes were then 
transferred to the adjacent Weipa minesite mosaic. Despite the 
50% cloud cover, several days difference in acquisition and 
References: 
Frame 2c 
5185 5168 
1576 1491 
6511 6370 
1410 1380 
  
differences in camera settings, the cloud-free areas of the Weipa 
mosaic were satisfactorily classified using those classes. 
Conclusion 
A heuristic approach has been adopted for correcting shading 
effects across DMSV images that involves correlation with a 
near-simultaneously acquired Landsat TM image. With this 
method 196 differently illuminated DMSV images, collected in 
1994, have been mosaicked and classified. Research is 
continuing to further refine the technique and to evaluate the 
effects of different solar angles and problems associated with 
the non-availability of near-simultaneously acquired TM data. 
Some breakdown of this method was detected at boundaries of 
areas with large brightness differences (e.g. between bare mine 
floors and native vegetation), where a slight blurring trend can 
be observed near the edges because of the large difference in 
the spatial resolutions of the two datasets. A method for 
reducing this effect is being analysed. However, the results are 
satisfactory for the intended use in minesite rehabilitation 
monitoring. 
King, D. (1991) Determination and reduction of cover type differential illuminations with view angle in airborne multispectral video 
imagery. Photogrammetric Engineering and Remote Sensing 57:1571-1577. 
Pickup,G., Chewings, V.H. and Pearce, G. (1995) Procedures for correcting high resolution airborne video imagery. [International 
Journal of Remote Sensing. Vol 16, No. 9, 1647-1662 
Lyon, R.J.P., Honey, F.R. and Hick, P.T. Second Generation Airborne Digital Multispectral Video: Evaluation of a DMSV for 
Environmental and Vegetation Assessment, Proceedings of the First International Airborne Remote Sensing Conference and 
Exhibition. Vol 2, 105-116 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996