The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
1344 
urban settlements of low density with limited vertical extent exist, 
in the study area, with a few number of water bodies. 
* 
31 Juni 
006 
0 i 
>2006 
. 
* 
• 
Figure 1. Location of images acquired for the 
Mausanne-les-Alpilles site. 
further fieldwork and acquire (using dual frequency GPS) a new 
series of 25 points at specific locations, chosen on the CartoSAT- 
1 imagery. The reference dataset comprises of a digital elevation 
model derived from airborne digital instrument with accuracy 
better than 0.60m at well-defined points. The reference DEM has 
a grid spacing of 2m, with standard projection, UTM 31N, and 
WGS84 datum. 
2.3 Ortho-rectification and DEM assessment 
The following approach has been adopted in this study; 
• 2-D accuracy of the given stereo pairs; 
• Estimation of the entropy to analyse the radiometry; 
• Image rectification and model adjustment; undertaking 
a series of ortho-rectification procedures with varying 
configurations of control points and check points; 
• Selection of the model and extraction of the DEM; and 
DEM quality assessment. 
• Ortho-image generation and checking the positional 
accuracy of the ortho-product. 
The black rectangle defines the nominal study area of interest. 
3. RESULTS 
1.3 Imagery acquired 
Two CartoSAT-1 stereo pairs are analysed and the basic 
parameters of the images are given in Table 1. 
Image 
Altitude 
Head 
Incid. 
Roll 
Pitch 
Yaw 
Al 
625.8 
194.0 
33.04 
-13.6 
-0.01 
2.25 
FI 
625.5 
194.0 
33.01 
-13.6 
-0.01 
2.10 
A2 
625.9 
194.1 
29.17 
4.01 
0.03 
2.70 
F2 
626.5 
194.1 
29.18 
4.01 
0.03 
2.56 
Table 1: CARTOSAT-1 stereo pairs 
Mausanne-les-Alpilles (MA/F1: 31Jan06, MA/F2: 06Feb06) 
A reference DEM of the test site TS-5 and ground control point 
coordinates, measured by GPS survey and corresponding image 
chips and photos of the measurement configurations, have been 
provided by the Principal Investigator (PI). 
3.1 Triangulation 
The RPC approach in LPS is used, applying the coefficients 
supplied with the imagery for triangulation of the two stereo-pairs. 
The model is fine tuned using 2 nd order polynomial adjustment. 
The performance of the model has been improved in both Easting 
and Northing directions. 
In the present study, 24 GCP’s are used for 2-D rectification and 
triangulation. 5 GCP’s with high residual error have been ignored. 
Different models have been adopted to rectify both the stereo 
pairs and assessed the accuracy with varying number of control 
points and checkpoints. 
Model-I GCP-19 
Model-II GCP-14 
Model - III GCP - 9 
Model-IV GCP-4 
Check Points - 0 
Check Points - 5 
Check Points - 10 
Check Points - 15 
The results of the above models are presented in the Table 2 and 
graphically represented in Figure 2. 
2. METHODS 
2.1 Software 
The main objective of the study is to determine the operational 
use of CARTOSAT-1 images using off-the-shelf software tools. 
Leica Photogrammetric Suite (LPS v9.0) is used for triangulation, 
DTM extraction and ortho-rectification processes. Analysis of 2- 
D geometry and radiometry is carried out using ERDAS Imagine 
software. 
2.2 Reference data 
Around 60 high quality (centimeter precision) GPS points are 
available for the test site. Due to the specific characteristics of the 
instrument under study, it is considered necessary to undertake 
Description 
Model 
I 
Model 
II 
Model 
III 
Model 
IV 
GCP - Ground X (m) 
2.27 
1.88 
1.56 
5.35 
GCP - Ground Y (m) 
1.49 
0.86 
2.14 
3.09 
GCP - Ground Z (m) 
2.39 
1.41 
3.41 
6.56 
GCP - Image X pixels) 
1.03 
0.90 
0.69 
1.75 
GCP - Image Y pixels) 
0.66 
0.44 
0.93 
1.18 
CP - Ground X (m) 
0.00 
3.54 
5.22 
10.74 
CP - Ground Y (m) 
0.00 
2.93 
1.84 
2.98 
CP - Ground Z (m) 
0.00 
6.27 
3.62 
11.20 
CP - Image X (pixels) 
0.00 
2.19 
2.52 
4.59 
CP - Image Y (pixels) 
0.00 
1.60 
0.82 
1.76 
Total Image RMSE pixels 
1.53 
2.02 
2.60 
5.63 
Table 2: Triangulation Results