1.63-9

	- [NEW] Added platform and Java information into the FITS header.
        This may make it easier in the future to identify platform-specific
        problems when they occur...

	- [BUG] Changing which pieces of pixel data (gains, flags, weights
	and offsets) could not be unset. This caused problems when the values
	stored in the FITS were corrupted, as there was no way of overriding
	the behaviour. This is now fixed. 

	- [IMPROVE] Made various tweaks to deconvolution and regridding
	algorithms. The updated deconvolution method now works independent of
	the sign of the source flux.

	- [FIX] Point source extraction tool 'detect' did not clear masks
	if these were supplied, leading to the masked sources being listed as
	detections. 

	- [FIX] Pressing 'u' in 'show' tool (for changing the image unit) did 
	not update the unit label. This has now been fixed.

	- [SPEED] improved performance of some old FFT related code. MiniCRUSH 
        benefits from these changes only when the 'spectrum' option is used.


1.63-8


	[NEW] Made crush image utilities compatible with the next generation
	image files. Starting from CRUSH version 2.0, the RMS plane will
	no longer be the raw observing rms, but will correspond directly to the
	noise in the signal plane, whether it is smoothed or not.
	This will make it much more convenient for users, who currently
	have to smooth the noise according to a recipe provided in the CRUSH
	document 'CRUSH-FITS-structure.pdf'.

	[FIX] Made conversion from pixelization to equivalent Gaussian
	smoothing consistent throughout.

	[COSMETIC] Minor changes to image information summary.

	[COSMETIC] 'show' tool has clickable components. You can change the
	colorscale, toggle between linear/log/sqrt scaling, or change units
	with just one click. You can drag the scale to adjust its range etc.
	The information panel provides you with hints.

	[COSMETIC] Added 'mJy/beam' unit for 'show' tool. Units can be changed
	by pressing the 'u' key, in the program window (or by clicking the
	the unit label on the right). 

	[COSMETIC] Updated 'show' utility to include coordinates, and used
	more human readable formatting. Also, it now has a cleaner look. 


1.63-7

	- [BUG] imagetool's '-origin' option resulted in botched y-coordinates.

	- [FIX] In locales where a decimal comma is used instead of a decimal
	point, the FITS headers became unreadable by other FITS tools. While
	this is strictly speaking a feature of the nom.tam.fits libraries
	used by crush, a workaround has been implemented.

	- [UPDATE] FITS libraries are recompiled.

1.63-6

	- [BUG] 'coadd' behaved as if all maps were centered on the same
	coordinate. This was due to an instantiation bug, by which all 
	map projections ended up being the same object, rather than separate 
	for each map. As a resust each map overwrite the projection information
	of the prior maps upon read.

	- [FIX] Removed unnecessary duplicate regridding step in 'coadd'.	

	- [FIX] 'JACKKNFE' keyword in coadded was always set to true. Now,
	it is fixed to correctly indicate if the map was coadded via random
	jackkniving...

	- [FIX] FITS libraries (nom.tam.fits) were not 1.5.0 compatible. They
        have been recompiled to restore 1.5 compatibility.

	- [CHANGE] Improved source extraction algorithm, in the way false
	detection probabilities are calculated. Also, the output now indicates
	the effective false detection rate instead of the Gaussian false
	detection rate before.

	- [UPDATE] Updated nom.tam.fits library to 1.00. This also fixes
	the FITS reading errors that resulted in some disabled functionality
	in version 1.63-3. Now all functionality should be restored...

1.63-5

	- [FIX] Restored lost backward compatibility to Java 1.5...

1.63-4

	- [FIX] Fixed error when trying to manipulate images w/o scan
	information.
		
	- [FIX] fixed S/N image sign when negative scaling factor is applied.

1.63-3

	- [UPDATE] Simplified FITS reading by making more use of common
	codelets...	

	- [CHANGE] Reverted back to 2nd order polynomials for faint reductions

	- [CHANGE] Points source corrections are now applied to all crush maps
	by default.

	- [CHANGE] Disabled acceleration response modeling (was used in 
	faint and deep reduction modes). There is no evidence that SHARC-2
	acts as an accelerometer :-)...

	- [CHANGE] Simplified SHARC-2 flux correction scheme by converting
	filter correction of models to equivalent Gaussian FWHM. 

	- [FIX] Adjusted the way the effective filter widths are calculated 
	when map is filtered multiple times. Unfortunately, in such cases the
	filter profile is no longer Gaussian. To first order, the width of
	the composite filter is mainly detemined by that of the narrowest 
	filter applied...

	- [FIX] re-writing of crush (SHARC2) images with imagetool was broken.
	It seems like the problem is with the FITS library itself... A
	workaround for now is that the scan info is discarded when the images
	are manipulated...

	- [FIX] Blanking seemed to have behavioural problems, causing funny
	flags appearing in the data. Besides, it did not even seem to do the
	job it was meant to. Thus, blanking is disabled for now by default.
	You can reenable it at your own risk...

	- [FIX] Pointing fits were broken.

	- [FIX] Several fixes to 'sourcecounts' functionality. 

	- [FIX] Filter corrections weren't always correctly calculated...

	- [FIX] imagetool's '-masks=match' could result in an error of the
	specified region fell outside of the map. Now the routine will return 
	a 'NaN' flux for such cases. 

1.63-2

	- [NEW] Added '-jackknife' option to 'coadd' allowing to generate
	randomly jackniffed maps. The FITS keyword 'JACKKNFE=T' is added
	to the image header to avoid confusion with properly coadded images.

	- [NEW] Imagetool has new feature '-growFlags' allowing to increase
	the flagged areas by a specified radius (or beam FWHM if no argument)
	is specified. This is useful to make sure that the areas surrounding
	a flagged pixels are also flagged. 

	- [NEW] Added extra column into source extraction output, indicating 
	the expected number of false detections above and including each 
	source.

	- [NEW] Updated 'histogram' tool to allow it to generate other
	than S/N histograms. The option '-image' can be used to specify which
	image plane to use for histogram generation (I.e. 'flux', 's2n', 'rms'
	'weight' or 'time'). Accordingly the value of the '-bin' option is
	interpreted to be in the default units for each image (I.e. the
	mapping unit for 'flux' and 'rms', its inverse square for 'weight', 
	seconds for 'time' and unitless for 's2n').

	- [BUG] Filter corrections were inconsistently applied when using
	extended structure filters (via FFT or convolution). Further
	inconsistencies between applying and undoing flux corrections. Both
	these problems are now fixed.

	- [CHANGE] 'K' in 'show' now always convolved to beam, whatever that
	may be... 

	- [FIX] An old workaround to make FITS double header values be readable
	by GILDAS software has been causing problems when decimals are marked
	with comma, rather than point in the locale. Thanks Ricky Nilsson for
	finding this one :-).

	- [FIX] Restored compatibility to Java 1.5. In a few instances 
	Arrays.copyOf function was used, which requires Java 1.6.


1.63-1

	- [BUG] extended filter resulted in unwanted unit change... 

	- [BUG] The regridding in the prior beta release (1.63-b1) was 
	buggy. This could produce faulty 'coadd' behaviour also. Now fixed.

	- [CHANGE] The '-crop' option in imagetool will now crop images
	automatically at the edges if no argument is specified. 

	- [CHANGE] Default FITS naming convention changed: instead of listing
	every individual scan that contributes to the map, only the first
	and last scan is listed when multiple scans are present. This results
	in much shorter, and human readable, default naming...

	- [CHANGE] Various improvements to coadd. When coadding directly 
	(default) the maps are loaded one-by-one, decreasing the memory 
	requirement. Regridding to common grid now happens only if the
	component maps do not already conform to one-another, and the 
	now happens in a single go, rather than the two-stage
	solution used before (i.e. first regridding, then shifting).

	- [CHANGE] Gaussian source fitting (e.g. for pointing reductions)
	has been overhauled to make it faster and more reliable.

	- [FIX] Region.add(FWHM) behaved like Region.addPoint()... This was
	no big issue since the function was not used thus far...

	- [FIX] 'coadd' now uses noise-weights measured on individual maps
	unless '-noise=data' option is specified, in which case the time-stream
	derived weights are used...

	- [FIX] Smoothing by a beam smaller than a pixel caused NaN errors.
	This is now fixed.

	- [FIX] Fixed glitch with 'imagetool -masks=match' overwriting the
	image data. In fact, the catalog matching should not output any image
	at all, rather it should exit when the catalog match is completed.

	- [FIX] Fixed FFT display issues...

	- [TUNE] Cleaned-up some of the SkyMap code...

1.63-b1

	- [NEW] imagetools's repertoir of operation extended to match sources
	to catalogs. The catalog can be loaded via the '-mask=<filename>'
	option, and the matching is performed via the '-masks=match' option.
	The output is the usual catalog/mask format, with 3 columns written
	into the comment field, in order:

		catalog S/N	peak pixel S/N	peak S/N

	The last two columns are the S/N levels that are found on the image
	for the given location inside a search area (which is calculated
	according to Kovacs 2006, assuming that the pointing rms <= the
	map pixelization.). 

	- [NEW] Introduced the ability to blank out bright source signals
	during reduction, s.t. their interference with the other models is
	minimized. This should help get rid of negative bowls around bright
	sources. The corresponding configuration key is MAP_BLANKING_LEVEL, 
	and command like option '-blank=', and it's value is a signal-to-noise
	level above which the blanking shoul be performed.
	Use together with -clip (MAP_CLIPPING_LEVEL), in similar way. 
	Blanking values >4 are recommended, to minimize its activity on noise.

	- [NEW] Introduced DEFAULT_ESTIMATOR key and '-estimator=" option
	to specify what type estimator is default. The accepeted values are
	'robust' and 'maximum-likelihood'. All else reverts to 
	maximum-likelihood. 

	- [BUG] Weighted median was (still) wrongly calculated. This is 
	hopefully fixed now. Problems could have resulted when median 
	estimators were used instead of the default maximum-likelihood ones.
	As a result of the fix, median estimators are now default for 
	signal estimation (but not for gains and weighting).

	- [BUG] Multiplexer modeling median calculation was buggy. This was
	not normally used however, so should not impact the average user...

	- [BUG] The source extraction tool 'detect' has applied filter 
	corrections twice when large-scale structure filtering was used (once
	at the time of filtering, and again at the extraction). Now any prior
	filtering flux corrections are undone first, to allow the extraction
	to operate on raw maps. The corrections are then reapplied to the 
	residuals at the end of the extraction routine.

	- [BUG] Pixel sizes were not individual to scans, rather they were
	set globally by the last scan read. This could have caused a 
	calibration problem of <10% when old Cassegrain mount scans were
	reduced together with Nasmyth mount scans. However, this caused
	no problem whatsoever with the reduction of like scans. 

	- [CHANGE] Source extraction tool 'detect' now estimates zero levels
	based on medians only in the first rounds. Afterwards it reverts to
	finding the mode (i.e. peak) of the distribution, which has almost
	no dependence on the shape of the noise distribution at large 
	deviations (thus it is not biased by the presence of sources or by
	the cleaning of noise).

	- [CHANGE] Source extraction tool 'detect' now estimates noise
	based on the negative side distribution alone. This should be more
	robust when a large number of sources are present in the map. Also
	more than one levelling/noise-restimation rounds are performed if
	necessary to uncover all sources. Improvements to internal flagging
	also.

	- [CHANGE] Reorganized source extraction algorithm, to select peaks
	in just a few rounds. This makes the routine perform at more or less
	constant speed irrespective of the number of sources (more likely it
	may be log(N)-type scaling...), rather than the old method, which 
	was roughly linear with source count N. This is a huge advantage 
	when using the method on a field with many sources.

	- [CHANGE] Added filtering information to displayed image info.

	- [CHANGE] Raw source maps produced with '-rounds=0' option will
	now at least zero-level timestreams.

	- [CHANGE] 'detect' no longer regrids input maps, since it has
	all the capabilities to work on coarse grids, which are also faster...

	- [CHANGE] Now supporting reading double-prescision BoA images.

	- [CHANGE] Slightly changed the quadratic interpolation method used
	for fine-tuning detection peaks in the source extraction.

	- [TUNE] Fixed gain iterations to 1 (instead of 'auto' convergent).

	- [TUNE] Reduced the number of iterations, as convergence is 
	achieved faster typically.

	- [TUNE] '-faint' and '-deep' reductions now fit only sky gradients
	not 2nd order polynomials

	- [TUNE] Removed some unnecessary costly re-calculations of
	SkyMap images...

	- [TUNE] Lots of stream-lining

	 - [FIX] In one place a Java feature sneaked in that is not avaiable
	before version 1.6.0, ruining backwards compatibility to 1.5.0 as
	advertized. The feature has now been rewritten to comply with the
	earlier API spec...

	- [FIX] Gain normalization is now always followed by an appropriate 
	renormalization of the correlated signals. In the past this was 
	effective only if direct gains were used, which was the default.

	- [FIX] The way the appropriate detection level was calculated for
	yielding the desired number of false detections (or confidence levels)
	had two issues: (1) it was an empirical formula based on random maps
	and (2) it was meant to be used with the old detection routine, which
	was pixelization dependent. The new routine has no pixelization
	dependence, and the number of independent points in a map can be
	determined (estimated) using Nyquist sampling theory. So this is
	the more correct approach for the current source-extraction method.

	- [FIX] Region flagging/unflagging was unnecessarily slow...

	- [FIX] When parsing masks files, lines that do not make valid 
	catalog entries are simply ignored. This makes it easier that plug
	catalog outputs (e.g. from 'detect') straight back into other CRUSH
	utilities (e.g. 'show').

	- [FIX] Improved source extraction after discriminative filtering:
	(a) by compensating for the effective blanking S/N change as a result
	of smoothing, and (b) by accounting for the filtering on the wings
	of the source, which fall below the filter blanking level.

	- [FIX] Speeded up source extraction by stream-lining region
	initialization.

	- [FIX] 'imagetool' and 'show' now both display image information
	after the manupulating steps.

	- [FIX] Various flagging fixes in source extraction. 
	
	- [FIX] Regions cleared after stacking. Without this, 'show' would
	display the circular regions (and/or polygons) inappropriately for
	the stacked image, when stacking is performed under 'show'.

	- [FIX] BoA map peak fluxes were slightly misinterpreted. They were 
	thought to have been normalized to instrument beams, instead they 
	were correctly expressed to image beams.

	- [FIX] NaN problem fixed when when trying to smooth map to the
	FWHM it already has...

	- [CLEANUP] Cleaned up some old code.

