An image taken by one of Maxar (DigitalGlobe) World View satellite cameras is formed of several blocks as tall as the image, mosaicked from left to right, with each block coming from an individual CCD sensor [Glo]. Either due to imperfections in the camera or in the subsequent processing, the image blocks are offset in respect to each other in both row and column directions by a subpixel amount. These so-called CCD boundary artifacts are not visible in the images but manifest themselves as discontinuities in the the DEMs obtained with ASP.
The tool named
wv_correct is able to significantly attenuate these
artifacts (see Fig. 5.2 in the
Digital Globe tutorial for an example). This tool should be used on raw
Digital Globe images before calling
It is important to note that both the positions of the CCD offsets and the offset amounts were determined empirically without knowledge of Digital Globe’s mosaicking process; this is why we are not able to remove these artifacts completely.
For PAN images, the WV01 and WV02 datasets are supported, for most TDI for the forward and reverse scan directions. For WV03 PAN images, CCD artifacts are less noticeable than for WV01 and WV02, and they are not corrected at this time.
For multispectral images, only a few select TDI are supported for band 3 of WV02 data. If a certain combination of spacecraft/TDI is not supported, the tool will print a warning and will write on output the uncorrected input image.
The ASP source code repository has additional documentation and tools for how to tabulate the corrections for the cases not yet covered by this tool.
16.69.1. Note for WV-2 images¶
Maxar (DigitalGlobe) WorldView-2 images with a processing (generation) date (rather than acquisition date) of May 26, 2022 or newer have much-reduced CCD artifacts, and for those this tool will in fact make the solution worse, not better.
ASP of version 3.3.0 or later (and any development build after 2023-06-21) will automatically detect this and will not apply the correction (a copy of the input image will be written on output). A warning will be printed in that case. This holds for both PAN and multi-spectral images.
This scenario does not apply to WorldView-1, 3, or GeoEye-1.
wv_correct [options] <input image> <input camera model> <output image>
Example for PAN images:
wv_correct pan.tif pan.xml pan_corr.tif
Example for multispectral images (first extract the third band):
gdal_translate -co TILED=YES -co COMPRESS=LZW \ -co BIGTIFF=IF_SAFER -b 3 ms.tif ms_b3.tif wv_correct --band 3 ms_b3.tif ms.xml ms_b3_corr.tif
Example if per-column corrections are available, for either PAN or multispectral images:
wv_correct --dx dx.txt --dy dy.txt image.tif image.xml image_corr.tif
16.69.4. Command-line options for wv_correct¶
- --ot <string (default: Float32)>
Output data type. Supported types: Byte, UInt16, Int16, UInt32, Int32, Float32. If the output type is a kind of integer, values are rounded and then clamped to the limits of that type.
- --band <integer (default: 0)>
For multi-spectral images, specify the band to correct. Required unless –dx and –dy are set.
- --dx <string (default: “”)>
For PAN or multi-spectral images, specify the plain text file having per-column corrections in the x direction, one per line, overriding the pre-computed table.
<string (default: “”)> As above, but for the y direction.
Print on standard output the per-column corrections about to apply (for multispectral images).
- --threads <integer (default: 0)>
Select the number of threads to use for each process. If 0, use the value in ~/.vwrc.
- --cache-size-mb <integer (default = 1024)>
Set the system cache size, in MB.
- --tile-size <integer (default: 256 256)>
Image tile size used for multi-threaded processing.
Tell GDAL to not create bigtiffs.
- --tif-compress <None|LZW|Deflate|Packbits (default: LZW)>
TIFF compression method.
- -v, --version
Display the version of software.
- -h, --help
Display this help message.