Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo

Close

image Log


Commit Date  
[a0c42a] by Carn? Draug Carn? Draug

Move conversion between each image type to imcast.

* im2double.m, im2int16.m, im2single.m, im2uint16.m, im2uint8.m: move code
into imcast. Fix several small bugs mainly dealing with indexed and logical
images, small precision for Matlab compatibility, and increased performance.
Expanded documentation. Added new tests.
* imcast.m: implement the conversion between each image type instead of being
a wrapper around functions for each conversion. This reduces code duplication,
and places all very similar (but not equal) code together so that a bug fix in
one can be easily notices that requires application in others. Expand
documetation. Add more tests.
* private/imconversion.m, private/im2float.m: remove no longer necessary
function since this has all been moved to imcast.
* COPYING: remove license for private/im2float.m and private/imconversion.m.
* NEWS: make note of bug fixes for this functions.

2014-03-24 22:00:05 Tree
[40269f] by Carn? Draug Carn? Draug

imcast: new function to convert image between arbitrary classes.

* imcast.m: new function acting as wrapper for im2double, im2single,
im2uint8, im2uint16, and im2int16.
* imcast.m, imnoise.m, imrotate.m: replace ugly hack using feval with
call to imcast.
* COPYING: specify license of new function.
* INDEX: add new function.
* NEWS: add note about new function for release 2.4.0.

2014-03-18 00:51:13 Tree
[1a31fb] by Carn? Draug Carn? Draug

fftconv2: clean up demo and block tests

2014-03-18 15:12:00 Tree
[0ed283] by Carn? Draug Carn? Draug

normxcorr2: expand for N dimensions, new tests and documentation.

* normxcorr2.m: use convn plus other changes so it accepts matrices or
images with an arbitrary number of dimensions. Expand documentation with
examples, and add tests for N dimensions, input checking, and precision.

2014-03-17 23:08:34 Tree
[e9c18b] by Benjamin Eltzner Benjamin Eltzner

normxcorr2: new Matlab compatible implementation (bug #41480)

* normxcorr2.m: complete rewrite of the function.
* NEWS: add note of new normxcorr2 and drop of dependency on the signal
package.
* DESCRIPTION: drop dependency on signal package (was only required for
xcorr2 (a, b, "coeff").

2014-03-17 21:19:48 Tree
[090388] by Carn? Draug Carn? Draug

maint: update version to development version 2.3.0+

2014-03-17 21:16:23 Tree
[4a96ec] by Thomas Weber Thomas Weber

doc: fix typos accomodate => accommodate.

2014-03-11 22:39:31 Tree
[43cd1c] by Carn? Draug Carn? Draug

bwdist: fix endless loop in bwdist with quasi-euclidean method in x86 systems.

* bwdist.cc: in a 32-bit machine, use of float instead of double was somehow
causing an endless loop (this may be treating the symptoms instead of the real
cause) in specific matrices. Added a test that duplicated this bug. This was
a regression caused by cset 27d87f06dd99 when changing the output class from
double to single for Matlab compatibility.

2014-03-09 20:41:08 Tree
[6ef0d8] by Carn? Draug Carn? Draug

maint: Periodic merge of stable to default.

2014-03-08 01:57:34 Tree
[72930d] by Carn? Draug Carn? Draug

Added tag release-2.2.1 for changeset fabc22044c6b

2014-03-08 01:12:06 Tree
[fabc22] (release-2.2.1) by Carn? Draug Carn? Draug

maint: release 2.2.1.

2014-03-08 01:11:47 Tree
[902e04] by Carn? Draug Carn? Draug

bwdist: some changes for performance only.

* bwdist.cc: make use of fortran_vec () when indexing the entire image
pixel by pixel, use boolNDArray for reduced memory and remove comparison
to 0.0, pass image by reference and use const, reduce number of OCTAVE_QUIT
macro.

2014-03-08 00:51:27 Tree
[9a7665] by Carn? Draug Carn? Draug

bwdist: fix regression calculating closest-pixel map (bug #41709)

* bwdist.cc: fix regression caused with cset 4bb352ade7c7. The original code
written in C (edtfunc.c file) had variable "h" for width and "w" for height.
This was probably because the code was originally written for row-major order
and it would be easier to swap these two variables than all instances of it in
the code. This fix swaps all instances of "h" and "w" in the code. Assign
number of rows and columns to "w" and "h" respectively would accomplish the
same but may confusion again in the future. New tests were added to check
the second output argument.
* NEWS: add notice of fixing this regression for next stable release.

2014-03-08 00:26:08 Tree
[70a4d2] by Carn? Draug Carn? Draug

bwdist.cc: make use of fortran_vec() in long loops for performance.

2014-03-06 21:31:41 Tree
[f1cf02] by Joonas Lipping Joonas Lipping

fspecial: weigth disk shaped filter by disk area covered (bug #41186)

* fspecial.m: change result of disk shaped filter for Matlab compatibility.
The value of each element corresponds to the area of the disk it covers. All
elements in the center will have the same value, but the border elements will
differ. Add new tests and update documentation.
* NEWS: add notice about the new behaviour.

2014-03-06 17:06:12 Tree
[714073] by Carn? Draug Carn? Draug

im2bw.m: display image instead of printint to stdout, when nargout is zero.

2014-02-18 23:10:42 Tree
[dac019] by Carn? Draug Carn? Draug

imcrop: silence assignment (missing ; at end of statement).

2014-02-18 22:49:22 Tree
[5e598c] by Jordi Guti?rrez Hermoso Jordi Guti?rrez Hermoso

regionprops.m: add test for cset b42bfcfed81d

2014-02-10 18:50:45 Tree
[b42bfc] by Jordi Guti?rrez Hermoso Jordi Guti?rrez Hermoso

regionprops.m: Return a column vector for Matlab compatibility

2014-02-10 15:41:30 Tree
[6f961c] by Carn? Draug Carn? Draug

Fix imrotate for RGB images, including input check regression on 660afd5cc1f4.

* imrotate.m: fix regression from 660afd5cc1f4 (error when ndims != 2) which
prevented rotation of RGB images, and replace calls to rot90() with rotdim()
which is used when angle is multiple of 90, so it works when there's multiple
dimensions (rot90 is limited to 2D matrices).
* NEWS: add this to the NEWS for the next patch release.

2014-02-02 23:41:13 Tree
[2d08c2] by Carn? Draug Carn? Draug

Compute complement of signed integers correctly.

* imcomplement.m: signed integers require handling different from unsigned
integers. Also, remove unecessary ismatrix check, add tests, expand on
documentation.
* NEWS: add this to the NEWS for the next patch release.

2014-01-31 21:09:12 Tree
[a17b71] by Carn? Draug Carn? Draug

Rewrite imcrop to add several alternative API's.

* imcrop.m: return rect (the position vector defining the bounding box), accept
indexed and ND images, use of current figure or specific figure handle, accept
specific rect argument for non-interactive usage, display cropped image if
nargout is 0, don't loop forever until valid bounding box is selected.
* NEWS: mention imcrop news for the next patch release.

2014-01-15 20:52:48 Tree
[dac212] by Carn? Draug Carn? Draug

Remove functions deprecated in release 2.2.0.

* bwborder.m, imrotate_Fourier.m, iptcheckstrs.m, impad.m, iptchecknargin.m,
readexif.m, uintlut.m: removed functions which were deprecated in 2.2.0.
* NEWS: list functions removed for next major release (2.4.0).

2014-01-14 18:39:28 Tree
[fa9074] by Carn? Draug Carn? Draug

Added tag release-2.2.0 for changeset b0977102e7f6

2014-01-08 13:29:23 Tree
[3ae7e2] by Carn? Draug Carn? Draug

maint: release 2.2.0.

2014-01-08 13:29:18 Tree
Older >