ISIS 3 Application Documentation
|Printer Friendly View | TOC | Home|
Checks the validity of a control network
This program checks for the validity of a Control Network. It is recommended to run cnetcheck before running jigsaw (can use the same input image cube list and control network) in order to catch a choice of potential issues with the control network and the corresponding input image cube list.
The default primary function of cnetcheck is to idenify any Islands within the control network (CNET). Each identified 'Island' will have an output file associated with it ("PREFIX"_Island.#, where #=sequential number id). The output Island files contain the list of images connected together as a group, the file includes the cube filenames and corresponding serial numbers.
NOTE1: If a serial number is identified as a single image island and the corresponding image cube file name is not listed within the FROMLIST, then the image cube filename will not be included in the appropriate island file.
NOTE2: If it is discovered that a known island is missing an output island file (i.e., output list not created as expected), it is possible that the image cube file name(s) were not listed in the input list (FROMLIST).
All output files are created in the working directory where cnetcheck is initiated.
A text file containing one column of ISIS3 image cube file names. Every image cube file used in the control network should be included in this list. The input images are required to be a Level0 or Level1 cube with a successful 'spiceinit' applied previously.
|Filter||*.txt *.lis *.list|
Use this parameter to specify the input Control Network file for validation.
This option determines which character will be used to separate the output file field entries, with the default option as a tab. Other options are comma, single space, or "OTHER," which allows the user to input whatever they wish as the delimiter character.
This parameter is only used if the user has selected "OTHER" as a delimiter. The user enters a string into this parameter to use as the output delimiter.
The results of the validity check are placed in output file names with the given prefix.
For example, PREFIX=CHK_, will result in an output filename CHK_Island.1 or CHK_NoControl.txt
Ignore=TRUE is the default for consistency with other apps, such as jigsaw, that do NOT consider invalid points or measures.
When Ignore=FALSE, all valid and invalid control points and control measures are considered for links between images and groups.
NOCONTROL=True, given the input image cube file name list (FROMLIST), any files (and their corresponding serial number) that are not found in the input control network (CNET) are considered images without control. These filenames and serial numbers will be listed in the output file named [PREFIX]NoControl.txt (with a field delimiter [DELIMIT]).
No output file will be created if all images are controlled.
NOCONTROL=False will turn off this testing of images without control.
NOCUBE=True, an output file named [PREFIX]NoCube.txt will be created, which will list all serial numbers encountered in the input Control Network (CNET) that do not have a corresponding image cube file name in the provided list (FROMLIST).
No output file will be created if all serial numbers match to an image cube file name.
NOCUBE=False will turn off the testing for matching serial number and input image file names.
SINGLEMEASURE=True, a file named [PREFIX]SinglePointCubes.txt will be created, which will list all image cube file names that exist as the single Control Measure in any Control Point within the provided Control Network.
Note: In general and when possible, a control point should have more than one control measures (more than three).
If there are no such single Control Measures, then no file is created.
When true, a file named [PREFIX]DuplicateMeasures.txt will be created, which will list all image cube filenames that exist in more than one Control Measure in a single Control Point.
Images under this scenario must be resolved before running jigsaw. A single image must not have more than one measure to the same control point.
If there are no such duplicate Control Measures, then no file is created.
When true, a file named [PREFIX]SingleCubes.txt will be created, that will list all image cube filenames that are measured to only one control point.
These images would have a convex hull value near zero, meaning there are not enough control points. Where possible, any image should have measures to multiple control points along the image plane boundaries (convex hull value near 1.0 for a full distribution). Refer to the Tolerance parameter for a distrubution test.
If there are no such single Cubes, then no file is created.
When true, a file named [PREFIX]NoLatLon.txt will be created, which will list all Serial Numbers within the given Control Network and FROMLIST which a latitude and longitude value cannot be computed at their respective image measure coordinate.
These identified Serial Numbers and image cube filenames will be followed by the Control Point IDs where the latitude and longitude could not be computed at the associated measures.
If there are no such serial numbers, then no file is created.
Only image cubes included in both the FROMLIST and CNET are checked.
The main purpose of this test is to isolate the images that have an insufficient distribution of control point measures. This is performed by constructing a convex hull from all Control Measures located on the image in "image pixel space". The convex hull gives a rough estimation of the distributed coverage of Control Points in the input control network across each image. By enabling this option, a file named [PREFIX]LowCoverage.txt will be created containing all images (with corresponding serial numbers) whose convex hull coverage is less than the given user-specified TOLERANCE. A third column will also be output, showing the actual coverage value for each reported image.
The default of 1.0 for a max tolerance will effectively cause nearly every image to be reported, as it is highly unlikely that a convex hull will cover 100% of it's corresponding image. With a tolerance of 0.25, any image whose convex hull covers less than a quarter of the image area will be reported. A minimal tolerance of 0.0 effectively makes the test meaningless, allowing images to "pass" this test without having ay control point coverage.
Only cubes in both the FROMLIST and CNET are checked.
Given the area of the convex hull, generated from all measures in an image, divide by the total area of the image. The resulting value from 0.0 to 1.0 shows roughly what percentage of the image is covered by its measures. If this coverage value is less than the tolerance specified by this parameter, then the image will fail the LOWCOVERAGE check and be reported as such. This value must be between 0.0 and 1.0, inclusive.