- » Geometric correction
Users of the NERC AZGCORR program may find the following notes helpful. They describe how to use AZGCORR to geometrically correct an airborne sensor image after having first atmospherically corrected the image. In its normal mode of operation, AZGCORR is applied to the original hdf data file, but most atmospheric correction programs require the data to be in a flat-file format (BIL or similar) and their output format is similar. This means that the atmospherically-corrected BIL file must be combined with the sensor/platform geometric information contained in the original hdf file to produce the desired result : an atmospherically-corrected image which may be overlain on a map in a GIS.
Geometric correction of the Thorney Island 24 July 2001 atmospherically-corrected data set using AZGCORR
The program that is used in this example to geometrically correct CASI images is AZGCORR, an Airborne Remote Sensing Geocorrection Package designed by Azimuth Systems and provided by NERC ARSF. The purpose of the program is to adjust an image data to match true ground space in a known coordinate system.
In order to apply a geometric correction to an atmospherically-corrected CASI image, the output image obtained from ACORN (c205021c_acorn) and the row image containing the HDF file (c205021b.hdf) have both to be imported into AZGCORR. The program will in fact read the information contained in the row image header file but apply the correction on the atmospherically-corrected image.
By typing the following command line is it possible to read the information contained in the HDF file:
azexhdf -h c205021b.hdf -v 13
The last parameter refers to the number of bands. This can be indicative if the exact number is not known; the program will then open the header file and provide the correct information.
In order to apply the geometric correction to the image, it is firstly necessary
to know the image pixel size and the number of lines defining the image. This
information is stored in the HDF file. The pixel size is calculated from the
flying height; in this particular example it is 10,200 ft and the pixel size
is therefore 6.2 x 6.2 for a CASI image.
A DEM grid of the area should also be added to the system, in order to improve geometric accuracy.
An example of a full command line applied to an atmospherically-corrected image is as follows:
azgcorr -1 c205021b.hdf -3 c205023c.hdf -p 6.2 6.2 -in -Bi c205021_acorn -B 1158 13 0 1.0 0.0 0 -e su6080.txt -ed 1 801 401 460000 100000 500000 120000 50 -mUK99 osgb99
The command -1 calls the input image, -3 the output, -p indicates the pixel size, -in changes the interpolation method from cubic convolution (default) to nearest neighbour, -Bi calls the BIL file, -B the image parameters (n. of lines, n. of bands, data type, scale, offset, fill value), -e calls the DEM grid while the last parameters refer to the area extent, altitude and map projection.
A full description of the different commands can be found in the AZGCORR
Once the image has been processed by AZGCORR, the azexhdf command is run again in order to obtain information on the new output image (c205023b.hdf). The data that are necessary for proceeding in the geocorrection process are contained under the header SCposimage. These are: no. of pixels, no. of lines, easting and northing coordinates and pixel size. AZGCORR outputs the easting and northing coordinates of the bottom right corner of the image. The image can now be exported and opened in ENVI.
When opening the image in ENVI, the option BSQ for byte interleave should be selected. The image has now to be saved as ENVI STANDARD format with a new name.
This new image is opened again in ENVI but to display it correctly the byte order has be changed to Network IEEE. This options is accessible by clicking on File>Edit ENVI header.
The image now requires to be rotated of 270° and transpose (Basic Tools>Rotate/Flip
Data) and the information contained in the header file further updated (Edit
ENVI header>Edit attributes>Map Info).
The coordinates of the upper left corner of the image must be entered as shown in Figure 1. This can be derived by applying the following formula to the northing value previously obtained from azexhdf processing:
(no. of lines-1) * pixel size + northing values
In this example, the result is:
(1185-1) * 6.2 + 99169 = 106.509.8
The resulting image is now geometrically correct. To test this, it is possible to overlay a vector dataset of the study area as shown in figure 2.
The image can now be saved as geoTIFF and imported in any GIS packages as shown in figure 3.