This is V2.2.2 (September  24 2001). New features include:
- Compiler-dependant neural filtering bug fixed.
- XMIN_IMAGE and co now follow the FITS convention (first
  pixel in image is at (1,1) in pixel coordinates).
- Fixed a typo in displayed text during background map-
  making

/*- Previous notes for V2.2.2 (February  8 2001) */
- Memory allocation bug for large images on Alphas while
  preparing the background-map fixed.
- CLEANing bug with hollow objects on weighted images
  fixed.

/*- Previous notes for V2.2.0 (February  4 2001) */
- MAP_RMS weight-map bug fixed.
- Huge MAP_WEIGHT calibration bugs fixed.

/*- Previous notes for V2.1.6 (February 11 2000) */
- crowding-flag bug introduced in V2.1.5 fixed.

/*- Previous notes for V2.1.5 (February 8 2000) */
- Small inaccuracy in crowding-flag positioning fixed.
- Fixed division by zero in local background in MASK_TYPE
  CORRECT mode.
- Risk of segmentation fault with huge weight maps fixed.

/*- Previous notes for V2.1.4 (November 13 1999) */
- Quote (') symbol now properly handled in FITS headers.
- Removed display of debug information inadvertently left
  in the previous release.

/*- Previous notes for V2.1.3 (September 14 1999) */
- Bug in ASSOC on Linux systems fixed.
- Background bug in WEIGHT_TYPE NONE,MAP_WEIGHT mode fixed.
- THRESH_TYPE ABSOLUTE now works on weighted images.

/*- Previous notes for V2.1.0 (May 24 1999) */
- Better formatting of output flags.
- New display.
- Total robust handling of weight-maps during background-
  determination.
- (tiny) differences between single and double-image results
  suppressed.
- New (experimental) _PROFILE photometric parameters.
- New WEIGHT_GAIN parameter (see doc).

/*- Previous notes for V2.0.22 (April 9 1999) */
- Many memory leaks fixed.
- CLEANing bug with very faint sources fixed.
- BACKGROUND_RMS check-images are now operational in
  single-image mode.
- Documentation improved.

/*- Previous notes for V2.0.21 (March 2 1999) */
This is V2.0.21 (March 2 1999). New features include:
- BACK_VALUE bug fixed.
- XPEAK and YPEAK offset bugs fixed.

/*- Previous notes for V2.0.20 (February 19 1999) */
- ASSOC bug in 2 ASSOC_PARAMS mode fixed.
- Variable thresholding is now taken into account when
  computing FWHM and CLASS_STAR if WEIGHT_TYPE != NONE

/*- Previous notes for V2.0.19 (January 13 1999) */
- SEGMENTATION numbering problem fixed.
- New Makefile macro-definitions.

/*- Previous notes for V2.0.18 (December 28 1998) */
- lint warnings fixed.
- new ASSOC_TYPE NEAREST option.
- MAP_RMS measurement mode and FLAG_TYPE MAX bugs fixed.
- new BACK_TYPE, BACK_VALUE and THRESH_TYPE parameters.
- FLUX_RADIUS does no longer require FLUX_AUTO.
- documentation slightly improved.

/*- Previous notes for V2.0.15 (August 20 1998) */
- ``Purify'' warnings and memory leaks fixed in several parts of the program.
- BLANKing bug of CLEANed pixels fixed.

/*- Previous notes for V2.0.14 (July 28 1998) */
- Internal handling of CDELTx changed (once more!) to allow proper scaling on the
  MINIBACKGROUND images.

/*- Previous notes for V2.0.13 (July 23 1998) */
- 2 small (but fatal on some machines) bug fixes.
- Comments (beginning with #) now allowed at every line in .param files.
- Default parameters updated in the config files (reflecting the increase
  of processing power during the last years!!)

/*- Previous notes for V2.0.11 (July 9 1998) */
- Background filtering bug at image borders fixed.
- CDx_x FITS WCS parameters are now given priority.

/*- Previous notes for V2.0.10 (June 30 1998) */
- Astrometry bug with CDx_x FITS parameters fixed.
- Interpolation bug in VAR_ONLY mode fixed.

/*- Previous notes for V2.0.8 (May 14 1998) */
- BACKGROUND check-image bug fixed.
- Old bug in filter normalization fixed.
- Bug in the reading of FITS header astrometric info fixed.

/*- Previous notes for V2.0.7 (May 3 1998) */
- Bug with VAR_ONLY interpolations fixed.
- Improve robustness toward incorrect FITS headers and empty images.
- A few bugs corrected (most of them introduced in the previous release!!)

/*- Previous notes for V2.0.5 (April 30 1998) */
- Multiple FLAG_MAPs now supported: IMAFLAG_ISO and NIMAFLAG_ISO have become
  vectors.
- Bug with FLAG_TYPEs MIN and MAX fixed.

/*- Previous notes for V2.0.2 (April 15 1998) */
- New banner
- Bug fixes: background on empty images, CLEANing in double-image mode.

/*- Previous notes for V2.0.0 (March 30 1998) */
- Renaming of the software and update of the proto-documentation
- New EXTRA_WARNINGS option to VERBOSE_TYPE config parameter.
- The usual amount of bug-fixes.

/*- Previous notes for V1.2b20b (March 24 1998) */
- XPEAK_ and YPEAK_ catalog parameters, that give the position of the brightest
  pixel in an object.
- THRESHOLD and MU_THRESHOLD are now set to the real extraction threshold.
- New CLEANing procedure.
- Many bugs corrected in the weighting/interpolation.
- SEGMENTATION check-image bug fixed.

/*- Previous notes for V1.2b18d (Feb 10 1998) */
- (Crude) interpolation with weight images, controlled with the INTERP set of prefs.
- BACKGROUND, MAP_VAR and MAP_RMS WEIGHT_TYPEs are now fully operational.
- Improved handling of bad pixels in photometry.
- First release of the first, very uncomplete "proto-doc" for the 2.0 version.

/*- Previous notes for V1.2b16h (Dec 15 1998) */
- Bug in end_writeobj(), bug in computing FLUXERR_ISOCOR, and bug with the pixel
  scales for non-WCS images fixed.
- New FLAG_TYPE config parameter for compositing the flag-map values (IMAFLAGS_ISO):
  OR, AND, MIN, MAX et MOST (keeps only the most frequent, non-zero flag-map value).
- New NIMAFLAGS_ISO catalog parameter, returning the number of selected flag-map pixels.
- Maximum string length for pathes, filenames, etc. extended to 512 characters.
- Handling of WCS CDxx_xx parameters (thanks to Doug Mink).
- WCS information propagated to MINIBACK-like check-images.

/*- Previous notes for V1.2b15 (November 4 1997) */
- Handling of environment variables within the configuration files.
- Selection of ASSOCiated data through ASSOCSELEC_TYPE: ALL means no filtering,
  MATCHED means keep only ASSOCiated data in the output catalog, -MATCHED is the
  opposite.

/*- Previous notes for V1.2b14 (October 26 1997) */
- Bug in the flagging cropped apertures fixed.
- BLANK and SCAN_ISOAPRATIO config parameters removed.
- New MASK_TYPE config parameter. Options are: NONE (no MASKing of the detections),
  BLANK (like BLANK Y in the previous versions), and CORRECT (photometric correction of 
  overlaps between objects by replacing with the symetrical pixel values)

/*- Previous notes for V1.2b13b (October 21 1997) */
- Handling of uncalibrated WEIGHT images for images with variable S/N:
  WEIGHT_IMAGE is the name of the external WEIGHT-map (FITS image)
  WEIGHT_TYPE parameter specifies which type of map it is (currently, only NONE
  and MAP_WEIGHT are supported). The detection threshold and CLEANing are now adaptive;
  photometric error computations take the variable S/N into account, too.
- DETECT_IMAGE parameter suppressed: in double-image mode, one should use instead
  of "sex <image>" "sex <detect_image>,<measure_image>"
- Bugs in the deblending routine and absolute threshold modes fixed.

/*- Previous notes for V1.2b10b (August 19 1997) */
- BLANKing option which allows masking of neighbour detections and makes some measurements
like MAG_AUTO or MAG_APER more robust towards contamination by other sources.
- FLAG_IMAGE option and FLAGIMA_ISO parameter for forwarding external flag-map information
to the output catalog.
- Support for multiple CHECKIMAGEs.
- New "-OBJECTS" CHECKIMAGE_TYPE: input image with both background and detections removed.
- Support for optional config parameters (like FLAG_IMAGE or DETECT_IMAGE)
- CLEAN_OBJSTACK has become MEMORY_OBJSTACK (reflects a deep change in the handling of
detections within SExtractor).

/*- Previous notes for V1.2b9b (June 8 1997) */
*Two bugs where found in V1.2b9a:
- X_IMAGE and Y_IMAGE where offset by 1 pixel if no WORLD info was requested
in the .param file.
- The DATE-OBS reading code was completely crippled. Thanks to B. Thomsen for the
info.
*New parameter:
- VIGNET_SHIFT(,) is similar to VIGNET(,), except that it produces a sinc-resampled
vignet with sub-pixel centering on object (beware of distorsions with undersampled
images).
*New ultrix Makefile option for DECstations running ULTRIX.

/*- Previous notes for V1.2b9a (June 1 1997) */
New features include:
- support for WCS coordinates. This means that you can get the
alpha and delta for objects detected on FITS images which comply with the WCS
standard (see http://www.cv.nrao.edu/fits/documents/wcs/wcs.html). The astrometric
mapping relies on Mark Calabretta's WCS library (V2.4 in C). SExtractor also
handles precession in the FK5 system, as well as transformations from J2000 (FK5)
to B1950 (FK4) coordinates. Shape parameters are transformed through a local,
linear approximation of the astrometric mapping. I am not an astrometrist, I
would therefore appreciate some feedback concerning these features. All
informations concerning precession in the FK4 system are welcome, as SExtractor
does only precess in the FK4 system presently. I would also gladly add support
for DSS images if someone can provide me with a decent description of their
astrometric parameters!
- 2D positional error estimate through ERRA_xxx, ERRB_xxx and ERRTHETA_xxx. Note:
these estimates are exact for unconvolved, white-noised images only.
- Linear photometric parameters (FLUX_xxx and FLUXERR_xxx in addition to MAG_xxx
and MAGER_xxx).
- Better outputs: units and vectors in FITS, (limited) backward compatibility
with the old V1.0 output format (CATALOG_TYPE FITS_1.0), new detailed ASCII
output mode (CATALOG_TYPE ASCII_HEAD).
- Handling of gigantic images (up to 2e9x2e9 pixels on 64bit machines).
- Online decompression of compressed DeNIS images.
- New "vignet" vector output: for instance VIGNET(8,5) in the .param file will
write in the output catalog a small 8x5 pixels array around each detection center.
- Many bugs corrected, among which: crash in the background mapping on
very large images and incorrect THRESHOLD output for blended objets.

The documentation is currently being written. In the meanwhile you may consult
the param.h file in the source/ directory for a minimum description of each
parameter.

/*- Previous notes for V1.2b5 (Jan 1997) */
Note that the CLASS_STAR, FWHM_IMAGE an FWHM_WORLD parameters haven't be tested
in this version (they shall be replaced by better estimators in future
versions). Many algorithms have been changed in this version. Therefore it is
expected that most measurements should yield results (slightly) different than
those obtained with earlier versions. In general accuracy and detection
performances should have increased, especially at low S/N; however I would
appreciate to have feedback from people involved in weak-lensing studies or
astrometrical work for instance concerning the accuracy of X,Y,A,B and THETA
parameters, which are now computed on FILTERED images.
The FITS catalog output format is yet in a temporary state; it is recommended
not to use it.
The documentation has not been updated yet; the following lines describe changes
made to the configuration files.

In the "config/" sub-directory you will find some configuration files. They are
basically the same as in the previous version except that:
- in the *.conv files:
  There is now a "NORM" keyword next to the "CONV" keyword. It just says that
  the convolution mask must be normalized ("NONORM" for no normalization). In
  previous versions this feature was controlled through the "CONVOLVE_NORM"
  keyword in the .sex file. However, old *.conv files are still accepted by the
  new version, and are automatically normalized, while a warning message is
  printed.
- in the default.sex file: many things have changed.
 *There is now a new "DETECT_IMAGE" parameter which requires a string as
  argument. If it set to "SAME", it means that the input image will be used for
  both detection and measurement, as in the old version. If you give it an
  image  filename (provided that the dimensions of this image are the same as
  those of the input image), it will be used as a kind of ``detection
  template'': all sources will be detected and deblended on this image, and will
  serve as reference for the measurements done on the regular image.
 *The EXTRACT_MINAREA keyword has become DETECT_MINAREA (for aesthetic reasons),
  but keeps exactly the same function. Idem for CONVOLVE -> FILTER and
  CONVOLVE_NAME -> FILTER_NAME.
 *The old THRESHOLD parameter has been now split in two: DETECT_THRESH and
  ANALYSIS_THRESH.
  DETECT_THRESH is the threshold of detection, just as before.
  You may express it in units of sigma of the background noise, or in surface
  brightness units, associated to a magnitude Zero-Point of the detection frame.
  Ex: "DETECT_THRESH 3.0" will set the threshold at 3.0 sigma,
      "DETECT_THRESH 22.0,25.4" will set the threshold at 22 mag.arcsec-2 for
      a detection-image whose mag-ZP is 25.4 mag. (note: the "PIXEL_SCALE"
      should have been set to the right value).
  In this way, the THRESHOLD_TYPE parameter has been eliminated.
  ANALYSIS_THRESH is a threshold in the measurement-frame (it can be the same as
  the detection-frame). It is used only for estimators that need accurate
  informations about the object profile in the measurement frame: star/galaxy
  separation and FWHM. If you don't know what to do with it, just leave it to
  3.0 (sigmas, you may also express it in mag.arcsec-2 like for DETECT_THRESH,
  but in this case the mag-ZP will refer to the measurement image).
 *BACK_XSIZE and BACK_YSIZE have been replaced by a single keyword,
  BACK_SIZE, which can accept up to 2 parameters (width and height of
  the background meshes). If only one number is given, width and height
  are simultaneously set to this value (useful because most of the time,
  people use square meshes!). The same happens for the BACK_FLTRXSIZE and
  BACK_FLTRYSIZE which have become BACK_FILTERSIZE. Similarly, the 3 Kron
  parameters PHOTOM_K... have been replaced by a common PHOT_AUTOPARAMS
  which requires 2 arguments, equivalent resp. to the former PHOTOM_KPAR
  and PHOTOM_KMINSIG. PHOTOM_KSIG is now set definitely to 6.0 in the
  code. This is because the new MAG_AUTO apertures are much more robust to
  crowding effects than before. Also, for aesthetic reasons,
  PHOTOM_APERTURE has become PHOT_APERTURES. There is an S at ApertureS,
  because the software can now photometer each source simultaneously
  through up to 4 apertures (see default.param). In default.param,
  MAG_APER is equivalent to MAG_APER1, and MAGERR_APER is equivalent to
  MAGERR_APER1.
 *CHECKIMAGE_TYPE now accepts a new type: MINIBACKGROUND. It produces in
  output a FITS image in which each physical pixel corresponds to one
  background mesh. This image is therefore roughly BACK_SIZE smaller than
  the input frame. The CONVOLVED option is now called FILTERED.
- The MAG_AUTO magnitudes are computed in a slightly different way than
  before, leading to a gain of about 20% in photometric precision in most
  extragalactic fields, especially in clusters. Obviously MAG_BEST
  benefits from this.
- The CLEAN_PARAM efficiency has increased, compared to previous versions.
  Smaller values (around 1.0) should now be used.

Please report all bugs to sextractor@iap.fr or bertin@iap.fr. Thanks!

						Emmanuel Bertin.
