[Insight-developers] Third quarter ITK release

Matt McCormick matt.mccormick at kitware.com
Thu Sep 20 18:02:09 EDT 2012


Hi all,

Thanks to funding from the NLM, we can start putting serious time into
an ITK release.  According to our release schedule, we targeted
mid-September for the final release of 4.3.0 , which has just passed.

There are import enhancement patches by Nick, Andriy, and others, but
there was relatively low activity on master this past cycle (which we
will hopefully fix with by the next release via increased community
contributions):

git shortlog --no-merges v4.2.0..master

Andriy Kot (1):
      ENH: Main filter for A2D2 PBNRR.

Arnaud Gelas (2):
      PERF: couple of performance improvements useful for mesh processing
      PERF: Minor performance improvement in KernelTransform

Baohua Wu (1):
      BUG: fixed a few bugs related to GPU Demons

Bill Lorensen (2):
      COMP: Macro definition of None in X11 breaks enum
      BUG: ThirdParty should not depend on ITK

Brad King (1):
      MetaIO: Avoid NULL dereference in MetaOutput::GetUsername

Bradley Lowekamp (8):
      COMP: fixing error when instatiated with image of double
      COMP: fixing error when instatiated with image of double
      PERF: remove per sample allocation, use per-thread structure
      BUG: prevent static initialization fiasco during factory registration
      PERF: Avoid memcpy call per component for Nift
      BUG: Correct for divide by zero when first bin has zero frequency
      PERF: add alignment data attribute to per-thread structure
      BUG: fix writing a corrupt meta image header from unknow metadata

Brian Helba (1):
      ENH: Update Remote/LesionSizingToolkit to newer commit

Cory Quammen (1):
      BUG: Added missing include <algorithm>

David Doria (3):
      ENH: Add a SetRegion function to all ImageIterators.
      ENH: Add SetRegion function to NeighborhoodIterators.
      ENH: Allow the ShapedNeighborhoodIterators to use SetRegion.

Dirk Padfield (1):
      PERF: Faster distance computation and fixed reflective iterator bug

Fotis Drakopoulos (1):
      BUG: Variance and structure tensor fix

Hans Johnson (19):
      ENH: Bump version to 4.3.0
      BUG: Test harness gave false positives
      BUG: Remove duplicate file in wrong place
      ENH: Allow removal of all Deprecated code
      COMP: Remove unnecessary const_cast.
      STYLE: Remove all references to StartRegistation
      BUG: Array overrun in LaplacianRecursiveGaussian
      COMP: Revert use of itkLegacyMacro
      ENH: Improve testing of GetInverse
      PERF:  Const correct optimize initialization.
      COMP: Fix data type for consistency
      ENH: Imaging data should not be in git
      COMP: Fix warning for propogating success status
      BUG: Missing DisplacementFieldTransform IO
      STYLE: Convert CMake-language commands lower case
      STYLE: Remove CMake-language block-end arguments
      BUG: Make IsTransformQueueEmpty const
      BUG: Reading composite transform error
      ENH: Update compiler fixes for fftw external

Ho Cheung (1):
      COMP: Add "#include <functional>" to classes for VS2012

ITK Migration V4 (2):
      DOC: Modify VectorIndexSelectionCastImageFilter comments.
      DOC: Modify ComposeImageFilter comments.

Kent Williams (7):
      PERF: Pull expensive GetInput/GetOutput calls out of inner loops
      COMP: propogate compiler selection & flags to FFTW build
      STYLE: Remove superfluous static_cast
      PERF: Pull expensive GetInput/GetOutput calls out of inner loops
      STYLE: Refine usage of static_cast
      BUG: B.Lowenkamp's changes to detect empty fields in Meta files
      BUG: isblank is missing on Microsoft Visual Studio.

Kris Zygmunt (2):
      PERF: Optimize patch-based denoiser tensor computations
      BUG: GPU tests sometimes fails

Matthew McCormick (19):
      ENH: Allow custom name for ProcessObject Primary Input/Output.
      DOC: Clarify ProcessObject NumberOfRequiredInputs.
      STYLE: Fix style in itkBlockMatchingImageFilterTest.
      BUG: Fix BlockMatchingImageFilter for pipeline updating.
      DOC: Use one column in the Doxygen alphabetical class index.
      BUG: Fix GCC_XML and SWIG ExternalProject dependencies.
      BUG: Fix GCC_XML and SWIG ExternalProject dependencies.
      BUG: Fix multi-object MetaIO reading for ASCII MetaVesselTube.
      STYLE: Improve readability and debugability in MetaScene.
      BUG: Fix multi-object MetaIO reading for ASCII MetaVesselTube.
      COMP: Fix QuadEdgeMesh doxygen warnings.
      ENH: Wrap new exponential Transform classes.
      STYLE: Cleanup WrapITK comments.
      BUG: Fix WrapITK GCCXML ExternalProject dependency.
      ENH: Use tarball for gccxml ExternalProject build.
      COMP: Make sure the VC script gets executed after gccxml installation.
      STYLE: Use typedefs in TreeNode.
      STYLE: Use more descriptive names in itkSceneSpatialObject.hxx.
      PERF: Use GetPrimaryOutput in ImageSource GetOutput.

Matthias Seise (1):
      COMP: Compile wrapITK for python2.7,64bit on windows

Michael Stauffer (3):
      PERF: Optimize GetValue in v4 image metrics.
      ENH: Add "WithOnlyIndex" iterators.
      PERF: improve speed of ImageToImageMetricv4 and derived metrics

Nick Tustison (12):
      ENH: Expanding HasLocalSupport to transform categories.
      ENH:  Adding ability to prepend transform queue.
      ENH: Adding labeled point set metric.
      ENH:  Add GetTransformCategory() to translation transform.
      ENH: Adding exponential transforms.
      ENH:  Adding common transforms to be registered.
      BUG:  Voxel size scaling needed
      ENH:  Enable inverse for exponential transforms.
      ENH:  BSpline automated scales estimation.
      BUG: Fix resampling of constant velocity field.
      BUG:  Uninitialized variable.
      ENH:  Reorganization of vel. field transforms.

Willi Huber (2):
      COMP: WrapITK generates non-unique typedefs in FFT submodule

And the dashboard is in need of repair:

  http://public.kitware.com/dashboard.php?name=itk

Instead of spending the next month in RC cycles with the repository
closed to enhancements, what are thoughts about one to two release
candidates for a v4.2.1 based off the current "release" branch?  There
are currently a number of fixes on "release" related to performance
and WrapITK.  Additional critical BUG and COMP patches could be added.


git shortlog --no-merges v4.2.0..origin/release

Bradley Lowekamp (1):
      COMP: fixing error when instatiated with image of double

Kent Williams (1):
      PERF: Pull expensive GetInput/GetOutput calls out of inner loops

Matthew McCormick (1):
      BUG: Fix GCC_XML and SWIG ExternalProject dependencies.

Willi Huber (1):
      COMP: WrapITK generates non-unique typedefs in FFT submodule


Thoughts?

Thanks,
Matt


More information about the Insight-developers mailing list