Orthorectification of Stereo Spot Panchromatic and Radarsat Fine Mode Data using Orbital Parameters
and Digital Elevation Model
Table 1: Characteristics of satellite data.
| Platform
|
SPOT
|
RADARSAT
|
| Scene Number
|
270-344
|
269-343
|
14297
|
C0015070
|
C0015059
|
| Sensor/ Mode
|
HRV-P
|
HRV-P
|
F5F-Mode
|
Standard
Mode (S2)
|
Standard
Mode (S7)
|
| Level of Processing
|
1 A
|
1 A
|
Path Image
|
Path Image
|
Path Image
|
| Date
|
21.4.98
|
23.1.98
|
31.7.98
|
20.12.99
|
17.12.99
|
| Pixel Size (meter)
|
10x10
|
10x10
|
6.25x6.25
|
12.5x12.5
|
12.5x12.5
|
| Viewing Angle (degree)
|
-30.3
|
+15.8
|
45.5
|
24
|
45
|
3. Data Processing
3.1 Data Collection
GCPs of high accuracy are necessary for rectification of SPOT Panchromatic and Radarsat Fine Mode data due to the high spatial accuracy of the data. For this reason, GCPs collection was carried out using GPS real time differential correction technique. The observations were done using a 12-channel Omni STAR receiver with sub-meter accuracy. A total of 35 GCPs that are uniformly distributed over the image were obtained. In addition, digitised contours from a topographic map with scale of 1:25,000 at 20 m interval are used to validate the DEM that has been generated. Major roads were also digitised to evaluate qualitatively the final orthoimage.
3.2 Geometric Modeling
In this study, the rectification process using PCI/OrthoEngine Version 6.3.0 software is based on collinearity conditions which represent the transformation between the image and ground space. This technique was developed by Toutin (1995) from the Canada Center for Remote Sensing (CCRS). The input for the model are orbital parameters gathered from header file and image coordinates (pixel, line) which correspond to the coordinates (X, Y, Z) of the Malaysian Rectified Skew Orthomorphic (RSO) projection.
The left-hand SPOT Panchromatic and Radarsat Standard Mode images were rectified. The rectified image is transformed and resampled to create epipolar geometry to ensure both images are offset only in the horizontal direction. Image matching was performed to match the corresponding pixel in the right image, i.e a template window and moving this template window in the search area of the epipolar image until the best digital number match is obtained. The correlation coefficient between 0 and 1 is calculated for each pixel, where 0 represents a total mismatch and 1 represents a perfect match. The different center locations of template window with the matched pixel represents the parallax. This parallax value is used to compute the elevation at the center of the template.
4. Rectification Results
4.1 SPOT Panchromatic Data
The rectification accuracy of scene 270-344 is 0.28 pixels (x direction) and 0.27 pixels (y direction) using all 35 GCPs. In order to find an optimum number of GCPs for the rectification, the following number of points were tested, i.e 4, 6, 8, 10 and 12 points. A few GCPs were also used as check points to evaluate the accuracy of rectification. Table 2 and Figure 3 show the
