| :: What's new AsterDTM 2.0 |
|
Coming Soon - What's new AsterDTM 2.2 - Please wait!
|
|
1. UPDATING AN EXISTING ASTERDTM INSTALLATION If you already have a valid entry
for AsterDTM in your current ENVI main menu file, you may update your
installation by executing the new AsterDTM installation file, clicking
"next" to go the following dialog box , and select "Only
update .sav file - no changes to menu.
|
| 2. OVERVIEW
|
| FUNCTIONAL ENHACEMENTS
There is a new option in AsterDTM for collecting and using Ground Control Points (GCPs), in order to either improve the georeferencing from ephemeris data and/or calibrate the elevation values to turn the relative output DTM into an absolute one. GCP's can be entered either in the form of a pre-existing GCP points file (ENVI default *.pts, through the "GCP Input file:" button), or by starting the collection of GCP's through the "collect GCP's" button in the start window. AsterDTM supports XYZ GCPs as well as horizontal (no elevation) and vertical (no coordinates) tie points.
AsterDTM now provides the possibility to produce orthocorrected and radiometrically corrected results from the Aster 1 A SWIR and TIR subsystems. This is a 2-step process: first the SWIR and TIR bands have to be corrected geometrically and radiometrically through the new "Create 1A -> 1B SWIR equivalent" and "Create 1A -> 1B TIR equivalent" options in the AsterDTM file menu; once these intermediate files have been produced , the SWIR and TIR bands will be available in the orthocorrection selection window. Geometric correction of SWIR and TIR 1A bands is achieved by automatic identification of tie points between the SWIR/TIR bands and VNIR band 3N, on a band-by-band basis , and subsequent registering. Once the operation has been completed, a check mark appears at the corresponding menu entry to signal the execution of this task. SWIR bands can optionally be passed through FFT filtering.
There is a new option "Output to radiance" now available in the orthocorrection dialog window for Aster 1B images, which converts the output images to radiance (Aster 1A images can be calibrated through the geometric and radiometric calibration).
AsterDTM can now automatically detect large water bodies like lakes, reservoirs, shores etc. , and assigns the elevation value of the border line to all water pixels. Water bodies (lakes, large rivers, sea) give a very low correlation signal and thus lead to spurious, random values in the output image. A new algorithm has been implememented which detects water bodies by shape, correlation values and image DN, and automatically assigns to these pixels the mean elevation value of it's boder/shore . The key parameters (minimum number of pixels per water body, correlation and image thresholds) can be modified through the processing parameters window. The automatic detection can be switched on through the processing parameter window and the shortcut menu.
|
|
NEW PROCESSING OPTIONS
There is now a new option available for outputting the DTM, correlation matrix and the orthocorrected images north oriented, instead of orbit oriented (which is the default). This option is accessible through the processing parameters window and the shortcut menu. Setting it will effectively rotate the output DTM and the correlation image so that they are north oriented, and set an option for north orientation in the orthocorrection dialog window. Having the output images north oriented is particularly useful when a mosaic of the output of several ASTER scenes has to be produced.
This new option allows the user to modify the kernel size for the image matching process. The correlation kernel is the image chip (subset) used to determine the offset between the nadir looking image and the backwards looking image, on a pixel-per pixel basis. Larger kernel sizes are more stable with respect to correlation analysis, but demand longer processing times and tend to 'blur" the output DTM. Smaller kernels lead to better resolutions for high contrast scenes, but may increase the noise and false matchings for more homogeneous images. The kernel size can be set through the parameter processing window, or the shortcut menu.
Image matching is a time consuming process, and is being executed in AsterDTM on several levels of pixel resolution in order to speed up processing. The last step executes on a relatively small range of possible elevations, which may not be sufficient when dealing with sudden terrain changes. Extended correlation analysis repeats the last image matching step in case the correlation for this pixel is low (< 60 on a scale from 0 to 100), but now with a much enlarged correlation window (possible elevation range ). The caveats of using the extended correlation are eventually a much longer processing time (mainly in the presence of water bodies or other low correlation areas), and a higher chance for false matchings.
The image matching algorithm looks for the optimum position (offset) of the correlation kernel on the image subset, using a linear cubic interpolation method to interpolate on a sub pixel basis., in the direction of the orbit orientation. Interpolation in addition in the direction perpendicular to the orbit orientation enhances the Z resolution slightly ( ~5 %), at the cost of a significantly higher processing time (about a factor of 2). AsterDTM now provides the option to select the bilinear cubic interpolation (2-D cubic) whenever desired, through the parameter processing window and the shortcut menu.
The processing of external DEM's has been completely reworked. External DEM's are now effectively interpreted as a collection of control points which fix the elevation at determined coordinates, thus providing support data where the correlation algorithm eventually fails (f.i. in the presence of clouds) . Output data points fixed by the external DEM are marked by value 101 in the correlation matrix. The offset between the external DEM and elevations derived with AsterDTM is being evaluated from high correlation data points, and applied to the whole image, thus effectively transforming the relative output DEM into an absolute one (the derived offset can be read out from the processing parameter window). The external DEM nodata value (typically -999, -9999, - 32735 etc.) is automatically defined from the input DEM file statistics. Support for SRTM data has been added.
The overall performance of the image matching algorithms has been improved through new adaptive scanning for high correlation internal tie points and enhanced interpolation of the correlation matrix. As a result, terrains with a high degree of variantion are delt with better now, and the vertical resolution of the output DTM has been improved to < 10 m RMS (CE60).
All of the processing options available for the interactive use of AsterDTM (with the exception of vector import and GCP support) are now available also in batch mode, most notably to provide an external DTM to improve the internal detection of tie points. For each image in the batch processing chain, AsterDTM verifies whether there is overlap with the provided external DEM and uses it if appropriate ( external DEM input may be a virtual mosaic, f.i. of SRTM data).
As of AsterDTM version 2.0, a short cut menu
is available with a subset of the processing options, accessible by clicking
with the right mouse button anywhere in the canvas of the AsterDTM start
window:
A check mark next to the options's text means that this option is currently selected. Please see the chapter "Options menu -> Defining processing options" for a detailed description of theses processing options.
The AsterDTM Installation and User's Guide has been completely revised to include all new features and enhancements of version 2.0 . In addition, tutorials and examples are being provided on the AsterDTM home page at www.envi.com.br/asterdtm to walk uses through the new functionalities.
|