EPIPOLAR

Name	Type	Caption	Length	Value range
MFILE *	String	Input file name, folder, or text file	1 - 192
DBIC	Integer	Input image channels or layers	0 - 4
MMSEG	Integer	Input math-model segment	0 - 1
DBIW	Integer	Raster-input window	0 - 4	XOff, YOff, XSize, YSize
SRCBGD	String	Source-background value	0 - 192	Default: FILE
SELMTHD	String	Pair selection method	0 - 192	PAIR \| MAXIMUM \| ALL \| OPT \| DENSE \| STRIPS \| AUTO \| PAIR_CONT Default: PAIR
MINPC	Integer	Minimum-overlap percentage	0 - 2	0 - 100
SAMPLING	Integer	Downsampling factor	0 - 1	Default: 1
OUTDIR	String	Output folder	0 - 192
OUTBGD	Float	Output-background value	0 - 1	Default: 0
TFILE	String	Text file name	0 - 192
GENOPT	String	Generation options	0 - 192
MEMSIZE	Integer	Working memory size (MB)	0 - 1	Default: 0
BREAKANG	Float	Flight line break angle	0 - 1

Parameter descriptions

MFILE

The name of a folder, image file, or text file that contains the raw input images to be processed. If necessary, you can use wildcards, such as the asterisk (*) in the value string.

You can specify the value of MFILE using any of the following:

A specific data-file name or folder path and file name; for example, "irvine.pix" or "C:\data\irvine.pix"
A specific file-name pattern or all of the files in a folder; for example, "C:\data\*.pix" or "C:\data"
An existing text file that contains references to multiple files with, optionally, different parameter settings on a per-file basis; for example, "C:\list.txt". The paths in the file can be absolute or relative to the folder location of the text file.

If the text-file option is used, the following general rules apply:

Specify a .txt extension for the file containing multiple files and different parameter values
List each set of parameters on its own line in the given order
Specify string parameters in quotation marks
Do not specify numeric parameters in quotation marks
A quotation mark preceded by a backslash is used as part of the string. For example, 'ab\"c' is used as: ab"c.
Precede any comments in the text file with an exclamation mark (!)
Use semicolons to delimit the parameters
If you do not want to specify a value, do not specify a value between the semicolons (";;"). Alternatively, if there are no remaining parameters to be specified on the end of a line, you can leave it blank.
Parameters in a text file may or may not be specified in the function:
- If a parameter is specified in the function and in the text file, the value provided in the text file is used.
- If a parameter is not specified in the text file, the parameter can be set in the function.

If you specify that value of MFILE as a text file, the files and parameters listed in the text file are used. For this task, the parameters in the text file must be delimited by semicolons and appear in the following order, where the DBIC, MMSEG, DBIW and SRCBGD parameters are optional:

[FILE];[DBIC];[MMSEG];[DBIW];[SRCBGD]

For example:

Format two lines in the text file as follows:

"C:\pci\Demo\Airphoto\S188.pix"; 1; 3;100,100,1000,1000;ALL,0

"C:\pci\Demo\Airphoto\S189.pix"; ; 3;;FILE,ANY,0

With S188.pix, only channel 1 will be processed. The math model will be read from segment 3. DBIW will be set to 100,100,1000,1000, and a SRCBGD rule of ALL,0 will be used.

With S189.pix, the parameters DBIC and DBIW will be used, but the math model will be read from segment 3 and a SRCBGD rule of FILE,ANY,0 will be used.

If the SELMTHD parameter is set to PAIR, FEPIPOLARCAND uses S188.pix as the first image and S189.pix as the second image.

DBIC

The input channel list of raw images used to create epipolar images.

If DBIC is specified in a text file (MFILE), this parameter is ignored.

If no value is specified for this parameter, all channels are processed by default.

MMSEG

The number of the segment in the input file that contains the math model for the raw images.

If no value is specified for this parameter, the last segment is used as the default.

DBIW

Specifies the raster window (x-offset, y-offset, x-size, y-size) of data to read from the input channels.

The same DBIW parameter is applied to all images, unless overridden by an entry in the MFILE.

If no value is specified for this parameter, the entire image is processed. XOff and YOff define the upper-left starting pixel coordinates of the window. XSize is the number of pixels that define the window width. YSize is the number of lines that define the window height.

Note: This parameter is particularly useful when generating epipolar images from scanned-film airphotos, as it can be used to exclude the registration marks and ancillary data printed along the edge of the image.

SRCBGD

Specifies which pixels in the source image are to be considered as background (NoData) pixels. In general, if a pixel is considered NoData, EPIPOLAR processes the pixel in a specific manner.

Available options are:

FILE: reads the NoData value from the input file's NO_DATA_VALUE metadata. The file-level metadata tag is used as the default for each channel; channel-level tags, when available, override the default. When metadata does not exist, each pixel in the input is considered valid.
0: each pixel, in each channel, having a value of 0 is considered NoData
NONE: each pixel, in each channel, is considered valid
<value>: each pixel, in each channel, having the specified value is considered NoData.
FILE [, <value>]: same as FILE, but defines the NoData value for each channel, when metadata does not exist.

For example, the default "FILE, 0" specifies that the function uses metadata. If the metadata is unavailable, pixels that have a value of zero are considered NoData>.

Note: To specify multiple values, use a comma-delimited list. The first value is applied to the first channel, the second value to the second channel, and so on. If fewer values are specified than the number of input channels, the last value is repeated for all remaining channels. If more values are specified than the number of input channels, the extra values are ignored.

SELMTHD

The method used to form the epipolar pairs. If no value is specified for this parameter, the Pair option is used by default.

Acceptable values are:

PAIR: automatically forms pairs according to the order in which the files are listed in the input MFILE or in the input folder. No other processing or screening of images is performed.
For example:
- File 1 uses the left image for pair 1
- File 2 uses the right image for pair 1
- File 3 uses the left image for pair 2
- File 4 uses the right image for pair 2
MAXIMUM: for each image, a single stereopair is generated using the image that provides the highest amount of overlap. Use the MINPC (Minimum Overlap Percent) parameter to specify the minimum percentage of overlap required to create a pair for consideration.
ALL: for each image, stereopairs are generated using all images that overlap above the minimum percentage specified in the MINPC (Minimum Overlap Percent) parameter.
OPT: stereopairs are generated to cover the entire project area, but an attempt is made to eliminate excessive duplication. The minimum-percentage value of the MINPC parameter is ignored with this method. Before pairs are selected, the images are assigned to contiguous flightlines, and only pairs of images from the same flightline are considered, to eliminate side overlaps. The OPT selection method is intended for use in processing airborne images, where the concept of flightlines is well defined.
DENSE: stereopairs are generated to cover the entire project area, with every image used exactly twice: as a left image and a right image of adjacent pairs. The minimum-percentage value of the MINPC parameter is ignored with this method. Before pairs are selected, the images are assigned to contiguous flightlines, and only pairs of images from the same flightline are considered, to eliminate side overlaps. The DENSE selection method is intended for use in processing airborne images, where the concept of flightlines is well defined.
STRIPS: The STRIPS selection method uses the same concept as the DENSE selection method. But it groups the images into flight lines using their geographic data extents. The STRIPS selection method will only give good results if the flight lines are straight and parallel to each other.
The STRIPS method allows up to two values for the MINPC parameter:
1. Along-Track-MINPC: this applies between images within a flight line. If ommitted the default value is: 60
2. Across-Track-MINPC: this applies between images between different flight lines.
  If you provide only one value for MINPC, then it applies to the Along-Track-MINPC only and the Across-Track-MINPC is set to 100, effectively turning off across track pairing. This behavior is to maintain backwards compatibility with how the STRIPS method worked before the two values were introduced.
AUTO: The AUTO selection method will use the best results from the STRIPS and DENSE selection methods. It will use the selection method that generated the fewest flight lines.
PAIR_CONT: automatically forms pairs according to the sequential order in which the files are listed in the MFILE parameter or in the input directory. The user can control the pair selection using minimum percentage overlap (MINPC). This method is intended for the automatic processing in the GXL system.
For example:
- Left file 1 uses the first file
- Right file 1 uses the second file
- Left file 2 uses the second file
- Right file 2 uses the third image

For the MAXIMUM, ALL, OPT, and DENSE methods, additional options may be specified in the form of KEY=value. Each KEY=value pair must be preceded by a comma.

The following KEYs and their values are recognized:

SSL: Stereo Separation Lower limit in degrees. Stereo separation is defined as the angle between the lines of sight from the centroid of the image-overlap area to the positions of the two cameras or satellites. If SSL is specified, the pairs with stereo separation lower than the provided value are not formed. If the SSL key is not provided or SSL or SSL= by itself is specified, the limit is set at 1 degree. The provided value must be greater than 0 and less than 180 degrees.
SSU: Stereo Separation Upper limit in degrees. If SSU is specified, the pairs with stereo separation higher than the provided value are not formed. If the SSU key is not provided or SSU or SSU= by itself is specified, the limit is set at 140 degrees. The provided value must be greater than 0 and less than 180 degrees. If both SSL and SSU are provided, SSL must be lower than SSU.
SAR: Specifies SAR-satellite-specific parameters to use in forming SAR epipolar pairs. The parameters are specified by single-letter codes, and can be provided as a contiguous string of letters that does not include the coma. The actual values are extracted from the metadata of each image. Candidate pairs are accepted if all specified SAR parameters match exactly; otherwise, the candidate pairs are rejected. If the SAR option is not provided, no SAR-specific screening is performed before forming epipolar pair candidates.
The following parameters (and the corresponding metadata tags) are recognized:
- N: Satellite platform name, given by the tag PlatformName
- L: SAR look direction, given by the tag LookDirection
- O: SAR orbit direction, given by the tag OrbitDirection
- B: SAR beam mode (without beam number), extracted from the tag BeamMode
- P: SAR polarizations, given by the tag Polarizations
- T: Product type, given by the tag ProductType
The SAR key is evaluated only for the MAXIMUM and ALL selection methods. SAR images without the required metadata tags are omitted from further analysis.

If SAR or SAR= by itself is specified, all parameters listed above must be identical for two images to form a pair (that is, the default is equivalent to SAR=NLOBPT). If only a partial list is specified, the parameters included in the list must match for the two images to form a pair, and the parameters not included in the list are ignored. Special forms SAR=ALL and SAR=NONE are also recognized. The NONE option disables all checks, but ensures that only SAR (as opposed to optical) images are used in forming pair candidates.

Note: The pairs accepted according to the criteria provided in the SAR key are still screened according to their stereo-separation criteria defined by the SSL and SSU keys. They must also satisfy the minimum-overlap percentage, as defined by the MINPC parameter.

MINPC

The minimum percentage value of overlap acceptable between two images for them to be considered a valid pair. If no value is specified for this parameter, the minimum-overlap percentage defaults to 50 percent.

The first step in the pairing is to evaluate every pair to determine if the overlap is greater or equal to the value of this parameter. For each image pair, A and B, two overlap values are calculated: A/B and B/A. Pairs are rejected when the minimum of A/B and B/A is less than the value MINPC / 100. The difference between the values A/B versus B/A is larger the more the two image sizes differ.

A/B is the area of A relative to the area of B
B/A is the area of B relative to the area of A

Image Overlap Calculations

Examples:

Example Overlap Scenarios

Stereo image sets - examples 1 and 2: when your data consists of stereo image sets, often forward and backward looking, you want only the stereo images to be paired together. In this case set MINPC to a high value, such as 75 or higher. For this example, the high value accepts the pairs AB and CD, and rejects the pairs AC, AD, BC, and BD.
Non-stereo image set - example 3: when your data consists of non-stereo image sets, similar to this example, you usually to pair images that are close to one another. In this case the default value value of MINPC is often good. For this example, it produces the pairs: AB, AC, BC, BD, and CD.
Non-stereo image set - example 4: for random images you want pair images might want to be less restrictive. In this case use a low value for MINPC, such as 40. For this example, it produces the pairs: AB, AC, and BD.
Strips - example 5: when the images are organized into flight lines (strips) with differing along-track and across-track overlap. In this case use STRIPS SELMTHD wioth two MINPC values, such as 60,20.

CAUTION:

This parameter is ignored with some values of the SELMTHD (pair selection method) parameter.

Note: The MINPC parameter allows for two values. The second value is only used with the SELMTHD STRIPS, and is ignored for other selection methods. See the STRIPS method to understand how the first and second values are used to find epipolar candidates.

SAMPLING

The downsampling interval for the created epipolar image. This parameter controls the number of pixels that are used from the raw images to generate the epipolar image. The cubic-convolution resampling is performed in converting the two images to epipolar geometry. Unless specified, one pixel is used for downsampling.

OUTDIR

The folder in which to create the output files. If no value is specified for this parameter, the output files are created in the current working folder.

OUTBGD

The background (NoData) value to use for epipolar pixels that are not populated. Each channel is set to the same background value. If the value specified for SAR images in decibel scaling is larger than -100.0, it is reset to -100.0. The provided background value will be truncated to the range allowed by the source-image data type.

TFILE

A text file, to which the following information is written:

[Epipolar_Left] [Epipolar_Right] [DBIC] [MMSEG]

The text-file format is delimited by semicolons and is in the same order as the MFILE parameter in the AUTODEM function. This text file can, therefore, be used as input for AUTODEM (MFILE = TFILE).

The FILE must be specified with the name and path of the generated epipolar image.

To name the output files:

Use the same name as the one used for the first input file; it must contain an underscore. The name of the second input file follows. Both files are prefixed with an el_ or er_ to indicate epipolar left or epipolar right, respectively. For example, S188 and S189 for epipolar left becomes el_S188_S189.pix.
The output folder is read from either the OUTDIR parameter in MFILE, from the task, or from the first input file. The raster channels created for each epipolar pair must be used for the DBIC parameter. For MMSEG, the math-model segment copied from the raw image to the output epipolar images must be used.

For example:

"C:\pci\EPI\el_S188_S189.pix";

"C:\pci\EPI\er_S188_S189.pix"; 1; 2

The paths in the file can be absolute or relative to the folder location of the text file.

If the text file already exists, it will be overwritten.

This parameter is optional.

GENOPT

The advanced options for creating epipolar images.

You can specify either of the following options:

ALL: Creates full-epipolar images, including any non-overlapping areas.
TRI: For use with tristereo imagery, such as Pleiades (forward, backward and nadir images).
Note: If you specify a text file for MFILE, the nadir file must be the last image listed.
CLIP=#: Option to specifiy percentage edge clipping from the original left image.

MEMSIZE

The amount, in megabytes (MB), of internal memory allocated to the task. The default value is 0 MB, meaning a value of 512 MB will be used on a 64-bit operating system (OS) and 128 MB on a 32-bit OS.

BREAKANG

Optionally specifies a flight line break angle, which is used to determine the end of a flight line. The vector angles between image centers are calculated , and if the vector angle between two image center is greater than the break angle threshold the flight line ends, and a new one begins. Default value is 22.5 dgrees.

CAUTION:

Use this parameter only if the value of the SELMTHD (pair selection method) parameter is OPT or DENSE.

Details

Epipolar images are pairs of raw images that are reprojected so that two images have a common orientation. Matching features between the images appear along a common x-axis. Epipolar pairs can be used in automatic extraction of a digital elevation model (DEM). Using epipolar images increases the speed of the correlation process and reduces the possibility of uncorrected matches.

The raw input files must have a math-model segment (MMSEG) before they can be used to create epipolar pairs. To obtain a math-model segment, export the model from CATALYST Professional OrthoEngine, or run CPMMSEG or OEMODEL.

The pair-selection method (SELMTHD) parameter allows you to select from a variety of pairing methods:

PAIR: automatically forms the pairs according to the order in which the files are listed in the MFILE parameter, or the order in which they are listed in the input folder.
For example:
- File 1 uses the left image for pair 1
- File 2 uses the right image for pair 1
- File 3 uses the left image for pair 2
- File 4 uses the right image for pair 2
With the PAIR method, the number of generated pairs will be half of the number of images selected for processing, or one less for an odd number of images. If the available images are A, B, C, D, E, F, G, H, and I, the generated pairs will be AB, CD, EF, and GH. This selection method is intended for internal use, as in some cases it may result in gaps in the stereo coverage of the project area.
MAXIMUM: automatically forms the pairs that demonstrate the highest amount of overlap. Use the MINPC parameter to specify the minimum percentage of overlap; pairs with overlaps below this percentage are not considered valid. With the MAXIMUM method, the number of generated pairs will be equal to or slightly less than the number of images selected for processing. If the available images are A, B, C, D, E, F, G, H, and I, the generated pairs will be AB, BC, CD, DE, EF, FG, GH, and HI, assuming all pairs meet the minimum-overlap criteria.
ALL automatically forms all pairs that overlap above the specified minimum percentage (MINPC). With the ALL method, the number of generated pairs may be several times greater than the number of images selected for processing, depending on the spatial layout of the image and the specified value of the minimum-percentage overlap (MINPC).
OPT automatically assigns images into spatially contiguous flightlines, and within each flightline selects pairs that provide full coverage of the flightline length, with maximum overlap within each pair. With the OPT method, the number of generated pairs will not exceed the number of images selected for processing, and in some cases may be less than half of the number. Assuming that the available images constitute two flightlines, with images A, B, C, D, E, and F in the first line, and G, H, I, and J in the second line, the generated pairs may be AB, DE, and EF for the first line, and GH and IJ for the second line. Note that the first and last pairs of every flightline are always generated, and no pairs are created from images belonging to different flightlines.
DENSE: automatically assigns images into spatially contiguous flightlines, and within each flightline selects all pairs with maximum overlap within each pair. With the DENSE method, the number of generated pairs will be equal to or slightly less than the number of images selected for processing. Assuming that the available images constitute two flightlines, with images A, B, C, D, E, and F in the first line, and G, H, I, and J in the second line, the generated pairs will be AB, BC, CD, DE, and EF for the first line, and GH, HI, and IJ for the second line. This method is similar to the MAXIMUM method, except the images are divided into flightlines and no pairs are created from images belonging to different flightlines.

The selection criteria can be further refined by the optional KEY=value pairs. The lower and upper values of stereo separation can be used for the MAXIMUM, ALL, OPT, and DENSE methods. For example, to restrict the DENSE pair-selection method to the range of convergence angles from 20 to 40 degrees, specify the value of the SELMTHD parameter as "DENSE,SSL=20,SSU=40".

The SAR key is specific to satellite-SAR images, and is evaluated only for the MAXIMUM and ALL methods. Typically, all of the parameters are used and, therefore, it is sufficient to specify the SAR keyword by itself, for example "MAXIMUM,SAR,SSL=5,SSU=20". The non-default settings should be used only in rare circumstances and for well-understood data, so that undesirable pairs are not formed.

For example, assume a combination of images is available from the same sensor, but with right-looking ascending orbits and left-looking descending orbits. The images can be processed with the SELMTHD parameter set to "MAXIMUM,SAR=NBPT". This will ignore the look-direction and orbit-direction values when forming the pairs, and accept all pairs with the difference in incidence angles of 1 degree of more. Note that for such pairs there may be a large difference in image orientations, and the extracted digital elevation model (DEM) may be of lower quality than expected.

In another example, assume images from two different satellites are available, and they are known to have identical polarizations and similar resolutions. Such data can be processed with the SELMTHD parameter set to "ALL,SAR=LOP,SSL=10,SSU=20". This will require the same look direction, orbit direction, and polarizations for both images, but ignore the sensor-specific platform name, beam mode, and product type. The pairs would be formed only from images with scene-center incidence angles separated by between 10 and 20 degrees.

With SAMPLING (Downsampling Factor) parameter, you can specify the number of image pixels and lines to use to calculate one epipolar-image pixel. For example, the value "2" indicates that one pixel in the epipolar images is formed by combining two adjoining pixels and two adjoining lines from the raw images.

Before processing, EPIPOLAR tests the convergence angle between images. Image pairs that exhibit poor stereo convergence will not be processed. The minimum acceptable convergence angle is 1 degree, and the maximum is 140 degrees.

Environments	PYTHON :: EASI :: MODELER
Quick links	Description :: Parameters :: Parameter descriptions :: Details :: Example :: Related

EPIPOLAR

Enhanced epipolar-image generation

Description

Parameters

Parameter descriptions

Details

Example