Last modified: December 2022

URL: https://cxc.cfa.harvard.edu/ciao/ahelp/vtpdetect.html
AHELP for CIAO 4.16

vtpdetect

Context: Tools::Detect

Synopsis

Voronoi Tessellation and Percolation (VTP) source detection

Syntax

vtpdetect  infile [expfile] outfile [regfile] [ellsigma] [scale]
[limit] [coarse] [maxiter] [edge] [superdo] [maxbkgflux] [mintotflux]
[maxtotflux] [mincutoff] [maxcutoff] [fittol] [fitstart] [clobber]
[verbose] [logfile]

Description

vtpdetect uses a Voronoi tessellation to locate connected sets of photons above a threshold density and group each set into a source. The tool analyzes the distribution of densities for significant source enhancements. An exposure time is assigned to each pixel location.

Strengths

Weaknesses

The "vtpdetect step-by-step" section at the end of this document contains a detailed description of how the tool identifies source detections. For a description of the theory and operation of vtpdetect, see the vtpdetect section of the CIAO Detect Manual and H. Ebeling and G. Wiednmann, Phys. Rev E, 47, 704 (1993).


Examples

Example 1

vtpdetect "my_evts.fits[STDEVT][pi=0:50]" my_exp_0_50.fits
my_srclist.fits

Read an event list filtered on the PI column from 0 to 50. Use the exposure map and output the source list to my_srclist.fits.

Example 2

vtpdetect "my_evts.fits[STDEVT][bin x=::32,y::32]" none output.fits

Bin an event list into an image, using a binning factor of 32.

For other vtpdetect examples, please see the CIAO Science Thread "Running vtpdetect".


Parameters

name type ftype def min max reqd
infile file input       yes
expfile file input none      
outfile file output       yes
regfile file   none      
ellsigma real   3      
scale real   1 0    
limit real   1e-06 0 1  
coarse integer   10 0    
maxiter integer   10 0 100  
edge integer   2 0    
superdo boolean   no      
maxbkgflux real   0.8 0    
mintotflux real   0.8 0    
maxtotflux real   2.6 0    
mincutoff real   1.2 0 10  
maxcutoff real   3 0 10  
fittol real   1e-06 0    
fitstart real   1.5 0.9 2  
clobber boolean   no      
verbose integer   0 0 5  
logfile file output stderr      

Detailed Parameter Descriptions

Parameter=infile (file required filetype=input)

Input file name (event list or image)

If the input file is an image, then the image is read in and a pseudo event list is created; events are added with a multiplicity equal to the pixel amplitude.

The tool is best run on a single chip at a time. vtpdetect assumes that the background is uniform across the field, so it will not work well on input files which contain both front-illuminated and back-illuminated ACIS chips.

Parameter=expfile (file filetype=input default=none)

There is preliminary support for exposure map files. This is assumed to be a 2 D image with WCS matching the input file.

Parameter=outfile (file required filetype=output)

Name of the output file.

Parameter=regfile (file default=none)

Filename for ASCII region output.

Parameter=ellsigma (real default=3)

Size, in sigmas, to make the elliptical source regions.

ellsigma is a multiplicative factor applied to sigma, the standard deviation of the distribution, to scale the major and minor axes of the ellipses for each source. ellsigma affects both the outfile and the ASCII region file (regfile). This feature is included so that the graphics overlay will be more visible and under the user's control. Often a value greater than 3 is helpful.

Parameter=scale (real default=1 min=0)

Background flux cutoff scale.

`vtpdetect' determines the background flux cutoff. This parameter allows the user to scale that value. Setting this to something greater than 1 will tend to de-blend sources while losing faint sources. Setting it to less than 1 will tend to blend source while picking out fainter sources.

Parameter=limit (real default=1e-06 min=0 max=1)

Number of false source events per number of background events.

Parameter=coarse (integer default=10 min=0)

Minimum number of events to consider to be a real source.

Parameter=maxiter (integer default=10 min=0 max=100)

Maximum number of iterations to allow.

Maximum number of iterations to allow. `vtpdetect' may converge before this. `vtpdetect' may also enter a state where the background estimate cycles between 2 or 3 values. Setting this to different values in this case will allow the user to control the exit level.

Parameter=edge (integer default=2 min=0)

The Voronoi tessellation at the edges is unbounded. This parameter controls how deep into the field to ignore events.

Parameter=superdo (boolean default=no)

Perform the super Voronoi cell update algorithm at the end of the program. Since this is a time consuming and uncalibrated routine this option is optional.

Parameter=maxbkgflux (real default=0.8 min=0)

Maximum normalized background flux level to fit with the cumulative Poisson distribution.

Parameter=mintotflux (real default=0.8 min=0)

Minimum normalized total flux level to fit with parabola.

Parameter=maxtotflux (real default=2.6 min=0)

Maximum normalized total flux level to fit with a parabola.

Parameter=mincutoff (real default=1.2 min=0 max=10)

Minimum absolute normalized flux level to find background cutoff.

Parameter=maxcutoff (real default=3 min=0 max=10)

Maximum absolute normalized flux level to find background cutoff.

Parameter=fittol (real default=1e-06 min=0)

Fit tolerance on Poisson fit.

Parameter=fitstart (real default=1.5 min=0.9 max=2)

Initial guess at Poisson fit.

Parameter=clobber (boolean default=no)

Remove output file if it already exists?

Parameter=verbose (integer default=0 min=0 max=5)

Level of informative output as the program is running. WARNING: verbose=4 and verbose=5 can produce hundreds of pages of output. Use only if needed to diagnose a problem during the percolation.

Parameter=logfile (file filetype=output default=stderr)

File where all messages will be directed. There are two special cases "STDOUT" and "STDERR" to indicate to use the C standard output (terminal) and standard error file.


vtpdetect step-by-step

vtpdetect locates connected sets of photons above a threshold density and groups each set into a source. Events are read from the event list and reduced to unique X,Y pixel locations (possibly with a "multiplicity" greater than 1, i.e. more than 1 event). An exposure time is assigned to each pixel location.

An area is assigned to each pixel location by computing the Delaunay triangulation from which the Voronoi tessellation is computed (the triangulation information is saved for the percolation step).

The cumulative distribution (cumulative density function, CDF) for non-source pixels is fitted on the low end (up to maxbkgflux) with a Poisson distribution. For the first pass, all pixels are considered to be background.

The minimum of the residual between the total (all pixel locations) CDF and the estimated background CDF (from above) is found between the mintotflux and maxtotflux ranges by two methods: absolute minimum and a fitted quadratic function.

A flux cutoff is determined from these two values and the values of the mincutoff and maxcutoff parameters and scaled by the user-input scale factor.

Below this flux cutoff, pixel locations are assumed to be background pixels. Above the flux cutoff, the pixels are grouped into sources based on the percolation (friends of friends) using the triangulation information to know which pixel locations are adjacent to every other pixel.

Given the total area of the source region (sum of the area of all pixel locations in source region), the minimum number of events needed to exceed the tolerated false source limit of background fluctuations is computed. If the number of events in the source (correcting for expected background) exceeds this minimum, then the source is significant. If not, then the pixel location with the lowest flux is eliminated from the source region and the percolation and significance testing is recomputed. (The percolation needs to be recomputed since the "trimmed" event may cause the source to split into more than 1 source region.)

After all the pixel locations have been searched and the sources found significant or not (they revert back to background events), the threshold is recomputed using newly found background events and the source list is recomputed from scratch. This process continues until the number of background events converges or maxiter iteration have occurred.

Super Voronoi Cell update: After the final iteration, the Voronoi tessellation is computed from just the set of source pixel locations and the edge pixel locations (to constrain the tessellation). To better account for flux in the wings of a source profile, the total flux is taken to be the total flux contained in these "super" Voronoi cells around each source. Then the background is estimated from this larger source area. (This option does not produce output other than to STDOUT at the end in lines that have "BEFORE" and "AFTER" comments.)

Finally the properties of the source regions are computed and output to the FITS source list.

Changes in CIAO 4.15

The spatial columns in the output table will now contain WCS transformation. This may mean that these virtual column names may be different from those created with previous versions of CIAO: rather than starting with EQSRC they will use the names of the WCS transformation in the input file.

Changes in CIAO 4.16


Bugs

The BKG_RATE and BKG_RATE_ERR columns are always equal to 1.0.

The values for the BKG_RATE and BKG_RATE_ERR columns are always set equal to 1.0.

The way the vtpdetect algorithm works the exposure corrected background rate is indeed constant for all sources. There is one background level above which Voronoi cells may be merged to create sources. The ESTBKG header keyword has the estimated number of background counts in the area covered by the tessellation, ie the TOT_AREA keyword.

The BKG_RATE and BKG_RATE_ERR column values are not used by vtpdetect.

Caveats

RA_ERR and DEC_ERR values near poles

The RA_ERR and DEC_ERR values are approximations and become increasing inaccurate the closer to the poles, DEC=+/-90.

Users may also see incorrect values for these errors for RA values very near 0|360 such that the error bar crosses the boundary and wraps around.

Workaround:

The X_ERR and Y_ERR values in physical pixels is correct. Users can use these together with the known plate scale to compute their own error estimates.

See Also

tools::background
create_bkg_map
tools::detect
celldetect, wavdetect, wrecon, wtransform
tools::gratings
tgdetect, tgidselectsrc, tgmatchsrc
tools::region
rank_roi, regphystocel, roi, splitroi