batcelldetect - perform source detection using the sliding cell method


batcelldetect infile outfile snrthresh


batcelldetect performs source detection on a sky image. The sliding cell method is used to locate regions of the image which are significantly different from the background.

This tool is more appropriate for coded aperture imaging because: (1) it assumes Gaussian fluctuations, not Poissonian; (2) it measures the local background; and (3) it measures the local background standard deviation. A source is detected at a pixel if that pixel's value exceeds the background by more than "snrthresh" times the background standard deviation. Users can increase the significance of a detection and reduce false detections by requiring more than one adjacent pixel exceed the threshold, using the "nadjpix" parameter. The "nullborder" and "bkgpcodethresh" parameters can also be used to exclude false positives near the edge of the image.

This tool performs analysis in two stages. In the first stage, new sources are detected using a sliding cell method. In the second stage, the fluxes and positions of both new and previously known sources are refined by fitting the point spread function.

In the first stage, the background and pixel significances are estimated by using a sliding window region method. The shape of the window is either a circular annulus or "square annulus," depending on the 'bkgwindowtype' parameter. The outer radius of the circle (or half-width of the square) is specified by the bkgradius parameter. The window region does not include the centermost portion (whose radius is the srcradius parameter). This portion should be the size of a typical source, so that the fluxes from sources do not contaminate the background estimate. For source detection, the pixel significance is estimated as the raw pixel value divided by the local standard deviation calculated using this method.

In a single iteration, the sliding cell algorithm is less sensitive in a region around bright sources, because the background standard deviation becomes biased. To avoid this, the algorithm can be run in multiple iterations (controlled by the 'niter' parameter). After each iteration, the detected pixels are removed, and thus the bias can be significantly reduced.

At the second stage, batcelldetect fits a point spread function to regions of the image where sources are detected. For new sources, this aids in refining the centroid of the source position, as well as in estimating uncertainties.

The default output is a catalog list of detected sources, plus various statistics about them. The output flux column is either COUNTS or RATE, depending on the input image units. The tool can optionally output the background map, the background fluctuation map, and a significance map.

IMPORTANT NOTE: if the keepbadsources parameter is 'YES', then the output catalog may contain some sources which failed processing for one reason or another. Information on the success or failure on a source-by-source basis is recorded in the DETECT_STATUS column.

The user can also supply an input catalog. Sources in the input catalog, which are within the field of view of the image, are assumed to be fixed at their known positions, and fitted during the PSF-fitting stage for intensity only (unless posfit=NO). Additional detected sources are also written to the output catalog.

This task can operate on multiple images, as produced by batfftimage. If the images vary only in their energy range, and are otherwise identical in their coordinate systems and exposures, then the 'vectorflux' parameter can be set to YES. When vectorflux=YES, each source has only one catalog row, and multiple fluxes are recorded in a vector table cell. When vectorflux=NO, a source will appear multiple times in the catalog, with a single flux in each entry. Note that if the input images do not match exactly their coordinate system or exposures, then you must specify vectorflux=NO.

To avoid spurious detections at the edge of the field of view, a partial coding threshold can be specified (see 'pcodefile' and 'pcodethresh'). If a partial coding map is provided, then by default sources must be at least 1% coded.

This routine uses a relatively straightforward brute-force convolution by FFT to compute the sliding-cell averages. Users should have a about a factor of eight more memory than the size of the image.

Fitting For Position and Flux

batcelldetect has several options to fit the positions and fluxes of sources. Newly detected and previously known sources are treated differently. By default, the positions of newly detected sources are allowed to vary and the those of previously known sources are held fixed. This behavior depends on the input parameters and the input catalog.

The following behaviors are possible:

Perform source detection only. For all sources, neither position nor flux are fit. For new sources, the positions are taken from the source detection stage. For previously known sources, the values are transferred from the input catalog completely unchanged.
srcfit=YES; posfit=NO (DEFAULT)
Hold the position of known sources fixed, but fit for the positions of new sources. The fluxes of all sources are fitted.
srcfit=YES; posfit=YES; posfitwindow=0.0
Fit for the positions of all sources. The positions of known sources are allowed to vary within a window which is 3 times the ERR_RAD value from the input catalog. Thus, if ERR_RAD = 0, the position is held fixed. Special cases: if an individual value of ERR_RAD is null, the position of that source is allowed to vary. If the ERR_RAD column is missing entirely, the positions are assumed to be fixed. The fluxes of all sources are fitted.
srcfit=YES; posfit=YES; posfitwindow > 0
As above, except the positions of known sources are allowed to vary by +/- 3*ERR_RAD or +/- posfitwindow, whichever is larger.
As of May 2007, the BAT team recommends using the GAUSSIAN point spread function (FWHM=0.37413) for all BAT data throughout the mission. The previous recommendation to use the "PYRAMID" shape was not correct.


The output catalog contains the following columns. Other input columns are copied to the output unchanged. For new sources entries, the row is filled with null value before writing batcelldetect-derived quantities.

Note that while the input catalog is only required to have the NAME, RA_OBJ and DEC_OBJ columns, they must have the units and dimensions listed below. Any input catalog columns that match the list below, must also have the specified units.

The data type should be either "D" (double) or "J" (integer); the NAME column can be up to a 30-character string. The word "Vector" indicates which output catalog columns will be a vector if vectorflux=YES and multiple images are analyzed.

Start time of image (Mission Elapsed Time). Type=D, Unit=s.
Stop time of image (Mission Elapsed Time). Type=D, Unit=s.
EXPOSURE time of image. Type=D, Unit=s.
Catalog identification number. For known sources, this number is copied from input catalog. For new sources, the number is determined by the 'newsrcind' parameter. Type=J.
Source name. For known sources, this name is copied from input catalog. For new sources, the name will be 'UNKNOWN'. Type=30A (string).
IMX Tangent plane coordinate (instrument X). Type=D.
IMY Tangent plane coordinate (instrument Y). Type=D.
IMX formal uncertainty. Type=D.
IMY formal uncertainty. Type=D.
IMX pixel coordinate (instrument X). FITS coordinate pixels label the center of the first pixel as IMXPIX=1.0. Type=D, Unit=pixel.
IMY pixel coordinate (instrument Y). FITS coordinate pixels label the center of the first pixel as IMYPIX=1.0. Type=D, Unit=pixel.
Source celestial longitude (equatorial or galactic). The presence of output celestial coordinates depends on the WCS coordinate system embedded in the input image. Type=D, Unit=deg.
Source celestial latitude (equatorial or galactic). Type=D, Unit=deg.
Source celestial longitude formal uncertainty (1 sigma). Type=D, Unit=deg.
Source celestial latitude formal uncertainty (1 sigma). Type=D, Unit=deg.
Estimated error radius (formal; 1 sigma). Type=D, Unit=deg.
Instrumental PSF X full width at half-max (tangent plane coords). Type=D.
Instrumental PSF Y full width at half-max (tangent plane coords). Type=D.
Instrumental PSF X FWHM formal uncertainty (1 sigma). Type=D.
Instrumental PSF Y FWHM formal uncertainty (1 sigma). Type=D.
Source angle from instrumental boresight. Type=D, Unit=deg.
Source instrumental longitude. Type=D, Unit=deg.
BAT simulator longitude. Type=D, Unit=deg.
BAT simulator latitude. Type=D, Unit=deg.
Background-subtracted source count rate (units from image BUNIT keyword)
RATE is used if the image has rate-like units. Type=D, Vector.
Source count rate formal uncertainty (1 sigma; units from image). Type=D, Vector.
Background-subtractd source count rate derived from the pixel nearest to the output catalog position. Type=D, Vector.
Estimated count rate due to all sources except this source; a measure of contamination due to diffuse background and nearby confusing sources. Type=D, Vector.
Background-subtracted source counts (units from image BUNIT keyword)
COUNTS is used if the image does not have rate-like units. Type=D, Vector.
Source counts formal uncertainty (1 sigma; units from image). Type=D, Vector.
Background-subtracted source counts derived from the pixel nearest to the output catalog position. Type=D, Vector.
Estimated counts due to all sources except this source; a measure of contamination due to diffuse background and nearby confusing sources. Type=D, Vector.
Local diffuse background intensity, ignoring any nearby point sources. This is the value used for background-subtraction of sources. The BKG column is filled with data from either the BKG_FIT or the BKG_CELL column. (units from image BUNIT keyword). Type=D, Vector.
Local background intensity formal uncertainty (1 sigma; units from image). Type=D, Vector.
Local background standard deviation (units from image). Type=D, Vector.
Local diffuse background intensity, as estimated using the PSF fitting technique. (units from image BUNIT keyword). Type=D, Vector.
Local diffuse background intensity, as estimated using the sliding cell technique. (units from image BUNIT keyword). Type=D, Vector.
Source signal to noise ratio (defined as RATE/BKG_VAR). Type=D, Vector.
Source signal to noise ratio (defined as CENT_RATE/BKG_VAR). Type=D, Vector.
Number of valid pixels in the source window. Type=J, Unit=pixel.
Number of valid pixels in the background window. Type=J, Unit=pixel.
Method used for source detection:
1=Cell detection
2=PSF fit.
4=Position calculated from multiple vector images
Chi2 value in PSF fit. Type=D, Vector.
Number of degrees of freedom in PSF fit. Type=D, Vector.
Reduced Chi2 value in PSF fit. Type=D, Vector.
Source detection status code; entries with non-zero DETECT_STATUS are deleted when keepbadsources='NO'; One of:
-1=Too many pixels detected;
-2=No pixels detected;
-3=Source touches border of image;
-4=Source error radius did not fit in source window.
-5=PSF fit failed (source was likely out of bounds)
PSF fitting constraint flag; Describes which parameters of the PSF model were fixed; Sum of:
1=IMX fixed;
2=IMY fixed;
4=COUNTS/RATE fixed;
8=PSF width fixed.
Source window radius . Type=J, Unit=pixel.
Background window radius . Type=J, Unit=pixel.
Number of good detectors (copied from input image). Type=J, Unit=pixel.
Near-field source height (ground testing only; coped from input image). Type=D, Unit=cm.
Partial coding fraction of source. Type=D.
Derivative of x position with respect to IMX. Type=D, Unit=deg.
Derivative of x position with respect to IMY. Type=D, Unit=deg.
Keywords copied from input image via the 'keepkeywords' parameter.


For the columns which refer to intensity --- namely RATE, COUNTS, CENT_RATE and/or CENT_COUNTS --- batcelldetect performs background subtraction. The source of the background estimate depends on the setting of the 'bkgfit' parameter. If 'bkgfit=NO', then the background is derived from the nearest pixel of the estimated background map. If 'bkgfit=YES', then the background is estimated during the PSF fitting process, as an additional constant level.

Whichever background method is chosen, the estimated background is subtracted from the raw intensity value before it is reported in the output catalog. The background value that was used is reported in the output catalog column named BKG. Hence, this equation holds for RATE, RAW_RATE = RATE + BKG, where RAW_RATE is the raw image peak height. In addition, the values derived from the two different methods are reported in the BKG_CELL and BKG_FIT catalog columns, corresponding to the background map and fitted PSF techniques respectively.

The background level is considered to be a contaminating source, and thus is added to the reported CONTAM_FLUX value.


Batcelldetect can correct for systematic non-linear centroid shifts in the image. It is assumed that all catalog celestial positions (input and output) are in "true" undistorted coordinates. All distortion corrections occur at the pixel level within batcelldetect. The distortion correction occurs if the 'distfile' parameter points to a correction map.


infile [filename]
Input file name containing one or more sky maps. Each sky map should be stored as one image extension. (Avoid using the CFITSIO syntax to select image extensions; individual images extensions can be selected from the file using the 'rows' parameter).

outfile [filename]
Output source list.

snrthresh [real]
Signal to noise threshold for detection of sources. A value of 6.0 indicates that an excess must be 6 sigma above the background level to be considered a detection. Here 1 sigma is the standard deviation of the background. Note: for performance reasons users should choose snrthresh > 3.5. Also, if nadjpix > 1, then the effective threshold will be higher than snrthresh.

(incatalog=NONE) [string]
Name of input catalog of known sources. This catalog must have at least the columns RA and DEC, in degrees (J2000). Additional columns are also copied to outfile (however see above about required units). A value of "NONE" indicates that no a priori catalog should be used. Specifying this incatalog=NONE by itself does not prevent the cell-detection based source detection algorithm from being applied.

(pcodefile=NONE) [string]
Name of optional partial coding map file. This may be any image with exactly the same dimensions and coordinate system as infile. Any region where the partial coding map exceeds pcodethresh is searched for sources; any region where it is below pcodethresh is ignored. The map need not be an actual partial coding map; any quantity that can be thresholded (even a map of 0s and 1s) can be used. If pcode file is "NONE", then the entire image is searched.

(pcodethresh=0.01) [real]
Partial coding threshold for source detection. This threshold is applied to the map specified in pcodefile. Any partial coding value greater than pcodethresh is considered active for the purposes of source detection.

(bkgpcodethresh=-1) [real]
Partial coding threshold for background analysis. This threshold is applied to the map specified in pcodefile. Any partial coding value greater than bkgpcodethresh is considered active for the purposes of measuring the backgroud and its variance. A value of -1 indicates that bkgpcodethresh should be set to pcodethresh. The "source" and "background" thresholds are kept separately so that background analysis can be done over a wider area than the source acceptance region.

(pospeaks=YES) [boolean]
Constrain newly detected peaks to be positve. If NO, then significant positive or negative deviations will be searched for. Known catalogged sources are permitted to have a negative fitted flux regardless of pospeaks.

(posfluxfit=NO) [boolean]
If YES, constrain the flux to be positive during the PSF fitting stage. If NO, fluxes are allowed to be positive or negative.

(vectorflux=NO) [boolean]
If YES, then output flux values for a given source as a vector, one element for each input image. If NO, then output one catalog rows for each source in each input image. Note, for vectorflux=YES, the input images must all match their coordinate systems, exposure, and instrument configuration exactly.

(rows = "-") [string]
An index list of images to operate on. This is a standard comma-separated list of image ranges, starting with image number 1. Image number 1 corresponds to the primary extension of the input FITS file, 2 to the next extension, and so on. A range of "-" indicates all images should be processed. Thus, this parameter really selects image extension numbers to operate on rather than binary table rows.

(niter=2) [integer]
Number of iterations of source detection to perform. niter=0 is allowed; in that case the sources in the input catalog (if any) which are in the field of view are transferred to the output catalog.

(carryover=YES) [boolean]
For the analysis of multiple images. If YES, then carry over the sources detected in the analysis of earlier images to the analysis of later images. Positions of previously detected new sources are held fixed as described above. If NO, then the analysis of each image begins from scratch, excluding any previously detected sources. Note that carryover=NO and vectorflux=YES are contradictory.
(nadjpix=3) [integer]
Minimum number of adjacent pixels required for source detection. This serves to decrease the number of false detections. This quantity is somewhat dependent on the image oversampling used.

(nullborder=NO) [boolean]
Allow sources which fall on the border of the image? If no, then sources whose "source window" overlaps with the edge of the image are ignored. The edge of the image is defined by the partial coding threshold. These sources may still appear in the output catalog with a DETECT_STATUS value of -3.

(bkgwindowtype="circle") [string]
Background window type, one of "circle", "smooth_circle" or "square". A "circle" is an annular background region with sharp inner and outer edges. A "smooth_circle" has the same shape but the edges are tapered in order to reduce noise fluctuations. A "square" is a square region with sharp inner and outer edges.

(bkgradius=30) [integer]
Background window radius in pixels. For a window of "circle", the bkgradius parameter is the radius of the circle. For a "square" window, the bkgradius is the half-width of the square. Actual diameter of circle or square is 2*bkgradius+1.

(regionfile=NONE) [string]
Optional name of source detection region file. Upon output, sources listed in the outfile catalog are also written to an ASCII "region" file, which can then be read into image display programs such as fv/POW or SAO ds9. Output units are degrees. The radius of the circle is twice the PSF gaussian sigma radius.

(srcradius=6) [integer]
Source radius in pixels. If 0, then the full background window is used to estimate the background and its standard deviation. Actual diameter of circle or square is 2*srcradius+1.

(srcdetect=YES) [boolean]
If srcdetect is set to "YES", then batcelldetect will attempt to detect new sources. If "NO", then source detection is disabled, but source fitting of previously cataloged sources may still be performed (depending on the value of 'srcfit').

(srcfit=YES) [boolean]
If srcfit is set to "YES", then batcelldetect will fit newly detected and previously cataloged sources (if any) as described above. If "NO", then previously determined positions and intensities are output. See "Fitting For Position and Flux" above.

(posfit=NO) [boolean]
Posfit specifies whether batcelldetect fits the position of already known sources. See "Fitting For Position and Flux" above. Also, the parameter 'posfitwindow' may override this parameter.

(bkgfit=YES) [boolean]
Fit the background for each source individually? If yes, then during the PSF fitting stage, each source has an independent background level. If no, a fixed background is subtracted.

(psfshape="GAUSSIAN") [string]
Shape of the point spread function to use. It should be one of "GAUSSIAN", "PYRAMID" or "TRUNCONE". "GAUSSIAN" is a two-dimensional gaussian function (which more closely matches the Swift/BAT point spread function in a single image); "PYRAMID" is a two-dimensional pyramidal frustum with a square base (no longer recommended); "TRUNCONE" is a truncated cone (no longer recommended).

(psffwhm="0.37413") [string]
Full-width at half maximum of the point spread function, in units of degrees. By default, this is held fixed, but if ":FIT" is appended, then the width is fitted. The default value listed above applies to the BAT point spread function.

(psftopwidth="0.04584") [string]
For the pyramidal point spread function, the full width of the top of the truncated pyramid, in units of degrees. By default, this is held fixed, but if ":FIT" is appended, then the width is fitted. The default value listed above applies to the BAT point spread function.

(newsrcname="UNKNOWN") [string]
Name to be used for newly detected sources. This can be a C format string, whose interpretation depends on the value of the newsrcind parameter. If newsrcind is greater than or equal to zero, then newsrcname may contain a single "%d"-style C format code. Each new source is given a unique source number, starting with index newsrcind. If newsrcind is -1, then newsrcname may contain two "%f" C format codes for the RA and DEC position, in degrees.

(newsrcind=1) [integer]
Starting index number to be used to label new sources (see newsrcname). If -1, then sources are labeled by their position on the sky.

(signifmap=NONE) [string]
Optional output file name for the significance map.

(bkgmap=NONE) [string]
Optional output file name for the mean background map.

(bkgvarmap=NONE) [string]
Optional output file name for the background fluctuation map.

(inbkgmap=NONE) [string]
Optional background map which overrides the internally-calculated background. The map or maps must have exactly the same structure and number of images as the input image file. If inbkgmap=ZERO, then the background is explicitly set to zero in the source detection stage.

(inbkgvarmap=NONE) [string]
Optional background variation map which overrides the internally-calculated variation map. The map or maps must have exactly the same structure and number of images as the input image file.

(distfile=CALDB) [string]
The 'distfile' parameter should point to a FITS file containing an MxNx2 image (or CALDB to use the default instrument correction in the calibration database). The two planes of the image are the non-linear distortion of the BAT "plate scale" as a function of position in the sky image. The values are offsets in tangent plane coordinates (IMX,IMY), and therefore unitless. The first plane of the image cube is the IMX offset, and the second plane is the IMY offset. The sense of the offset is (TRUE-APPARENT), i.e. for a measured position in tangent plane coordinates, the offset values should be *added* to arrive at the true position in tangent plane coordinates. The images are low-resolution versions of the BAT field of view in instrumental sky coordinates, and are meant to be interpolated to the desired sampling. The WCS keywords must specify the image coordinate systems.

(keepbits = "ALL") [string]
Integer number of bits of precision to keep in the output images. A value of ALL means keep all bits. A positive value indicates that 'keepbits' bits of precision should be preserved, and the other bits should be set to zero. Fewer kept bits means the output images are more compressible, at the expense of lost precision. For background and variance maps, a straight number of bits are kept (thus preserving the same relative precision across the map). For significance maps, 'keepbits' prescribes the number of noise bits to keep; more bits may be kept if a pixel is significantly above the noise (thus preserving the same noise quantization error across the map).

(npixthresh=20) [integer]
Minimum number of pixels in the background window to enable source detection. Setting this parameter to a high number will prevent detections based on only a few background pixels.

(posfitwindow=0) [real]
The position 'window' radius, in degrees, used for all position fits, including known sources. The window is the allowed region over which the centroid position is allowed to vary during the position fit, regardless of the error radius specified for the source. Typically, this parameter will only be used if the attitude solution is suspected to be incorrect. A value of zero indicates that the window size is determined by the error radius specified in the catalog, if any. WARNING: This parameter is basically an override for the ERR_RAD column and the 'posfit' parameter. See "Fitting For Position and Flux" above.

(possyserr=0) [real]
Systematic position error, in degrees, to be added in quadrature to each position error radius reported in the output catalog. This parameter allows one to specify a minimum systematic position error for every reported detection (which is separate from posfitwindow, which controls the position determination process itself).

(vectorposmeth=AVERAGE) [string]
Specify which source position is reported when the source position is fitted and vectorflux=YES. The output catalog has only one position for each source. If position fitting is allowed and vectorflux=YES, then it is possible for multiple positions to be derived. This parameter describes what to do with multiple positions. The allowed values are: "FIRST" or "FIXFIRST" - report the fitted position in the first image; "LAST" - report the fitted position in the last image; "AVERAGE" - report the weighted average fitted position; "MAX_SNR" - report the fitted position for the image with highest signal-to-noise ratio. "FIXFIRST" is different than "FIRST" in that once a fitted source position is found, its position is held fixed for subsequent images. If the fitted position does not exceed 'snrthresh' then it will not be included in these calculations. When positions are combined using one of the above methods, then the DETECT_METHOD column will be flagged with a multiple of 4.

(keepkeywords="*APP,CLEANLEV") [string]
A comma separated list of keywords or keyword wildcard patterns. The matching keywords are transferred to the output catalog, under a column of the same name. This is a way to automatically keep track of corrections that have been applied to the input image. The CFITSIO wildcard patterns are: '?' will match any single character at that position in the keyword name; '*' will match any length (including zero) string of characters; and '#' will match any consecutive string of decimal digits (0 - 9).

List of comma-separated image type filters, used to accept or reject images based on their HDUCLASn keywords. This is useful for files that contain different kinds of image extensions. A minus sign "-" indicates exclusion. Each image is checked for each of the HDUCLASn keywords. If any of the keyword values matches an excluded list member, then that image is rejected for analysis. List members without a minus sign must be present. A value of NONE indicates no filtering (i.e. all images are accepted).

(sortcolumns="-KNOWN,-SNR/UNKNOWN,RA_OBJ") [string]
Controls the sort order of the table displayed on the console. Note that this does not affect the FITS table, which can be sorted using ftsort or fsort. The string should be a comma-separated list of sort keys, or "NONE" for arbitrary printed order. The first key listed is the primary sort key, the next is the secondary and so on. By default, values are printed from lowest to highest, but a preceding '-' sign indicates highest to lowest. The following sort keys are available: NAME; RA_OBJ; DEC_OBJ; SNR; RATE; COUNTS; FLUX (all three of these keys are synonymous); KNOWN (if the sources appears in the input catalog or not); IMX; IMY; PCODEFR; CATNUM; THETA; PHI; SNR/UNKNOWN. Most sort keys are identical to their corresponding catalog column, described above. "SNR/UNKNOWN" is a special sort key, indicating to sort only previously unknown columns by SNR. The default sortcolumns indicate to arrange the table by printing the known sources first, sorted by RA, and then the unknown sources, arranged in descending order by signal to noise ratio.

(keepbadsources = NO) [boolean]
If yes, then individual sources are kept in the output catalog, even if DETECT_STATUS is non-zero, indicating a failure of some kind. If no, then any output catalog entry with DETECT_STATUS non-zero is removed.

(ptcorrel = "NONE") [string]
Account for pixel to pixel correlations during the fit. Either NONE (straight PSF fitting); PSF (use PSF for correlations); or GAUSSIAN[x] (where x is the FWHM of the correlation function in degrees). NOTE: This parameter is not tested and may produce unexpected results!

(clobber = NO) [boolean]
If the output file already exists, then setting "clobber = yes" will cause it to be overwritten.

(chatter = 2) [integer, 0 - 5]
Controls the amount of informative text written to standard output. Setting chatter = 1 produces a basic summary, including a list of detected sources; chatter = 2 (default) additionally prints a summary of input parameters; chatter = 5 prints debugging information.

(history = YES) [boolean]
If history = YES, then a set of HISTORY keywords will be written to the header of the specified HDU in the output file to record the value of all the task parameters that were used to produce the output file.


1. Performs source detection on skyimage.img, finding those sources which exceed the background by 6 sigma. The output is written to srclist.fits. The default circular background window is used.

        batcelldetect skyimage.img srclist.fits 6.0

2. Same as example 1, but use a square background window of half-width 10 pixels.

        batcelldetect skyimage.img srclist.fits 6.0  \
            bkgwindowtype=square bkgradius=10

3. Same as example 1, but outputs the significance image to signif.img as well.

        batcelldetect skyimage.img srclist.fits 6.0 signifmap=signif.img




The fitting library, MPFIT, is based on MINPACK. This product includes software developed by the University of Chicago, as Operator of Argonne National Laboratory. See 'fhelp batcelldetect-mpfit' for more information.


Oct 2009