Changes DIPlib 3.2.0

Changes to DIPlib

New functionality

• Added dip::BlendBandlimitedMask().

• Added dip::FreeTypeTool to render text into images. It depends on the FreeType library to use any font file for rendering the text. This functionality is opt-in (that is, by default it is not enabled).

• Added dip::DrawText() to render text into images using a built-in, fixed-size font.

• dip::Image::Pad() can now pad with an arbitrary pixel value.

• dip::Histogram now has a stream output operator (<<), which shows basic information about the histogram.

• Added dip::ObjectMaximum() and dip::ObjectMinimum() to identify the object with the largest or smallest measurement result.

• Added dip::Image::UnexpandSingletonDimension() and dip::Image::UnexpandSingletonTensor().

• Added dml::IsString() to the DIPlib–MATLAB interface.

Changed functionality

• The output stream operator for dip::Image writes the same information in a slightly different way.

• dip::Histogram::Bin( FloatArray const& value ) now takes the input as a const reference.

• The scan framework will preserve expanded singleton dimensions: if one dimension is singleton-expanded across all input images, then the output images will have the same dimension be singleton-expanded. See issue #96.

• The dip::libraryInformation struct now also shows whether FFTW and FreeType are being used.

Bug fixes

• When inserting a dip::Measurement object into a stream, the table it produced sometimes had feature value columns narrower than the feature name header.

• Fixed addition and subtraction operator for dip::Histogram changing the LHS operand: h3 = h1 + h2 would end up with h1 equal to h3.

• Fixed bug that prevented compilation with MinGW. See issue #90.

• The documentation was missing a lot of functions in the arithmetic, comparison and trigonometric operators modules, as well as the macros in the DIPlib-MATLAB interface.

• dip::Distribution::Y and dip::Distribution::Ybegin didn’t throw an exception when attempting to access one past the last y value.

• dip::Distribution now defaults to no x-units (rather than pixels), making dip::PerObjectHistogram correctly not have units for the x-axis, which is intensity, not pixels.

• dip::GreyWeightedDistanceTransform didn’t properly use pixel sizes as intended with the default "fast marching" method.

• Updated DocTest to version 2.4.8, which should allow building and running the tests on Apple Silicon devices.

Changes to DIPimage

New functionality

• Added drawtext function.

Changed functionality

• In the DIPimage GUI, “About DIPimage” and “About DIPlib” show the same URL (using HTTPS) as a link. The “About DIPlib” page also reports whether FFTW is used.

• readim and writeim recognize 'NPY' as a format. This is the NumPy array format supported natively by DIPlib.

• bdilation, berosion, bclosing and bopening have a new option 'isotropic'.

(See also changes to DIPlib.)

Bug fixes

• Fixed build script to allow building with newest version of MATLAB. See issue #88.

• Updated dipmex to work properly with the MinGW compiler on Windows, and with older versions of MATLAB.

(See also bugfixes to DIPlib.)

Changes to PyDIP

New functionality

• Added Python 3.10 to the list of deployed wheels.

• Added dip.Histogram as a class, matching DIPlib’s dip::Histogram. This gives a lot more flexibility when creating the histogram and when working with histograms. Most of the class methods from DIPlib have been made accessible from Python, as have all the histogram processing and analysis functions:
  • dip.Histogram.Configuration class.
  • New dip.Histogram class methods: IsInitialized(), Copy(), ReverseLookup(), Dimensionality(), Bins(), BinSize(), LowerBound(), UpperBound(), BinBoundaries(), BinCenters(), BinCenter(), Bin(), GetImage(), Count(), Cumulative(), GetMarginal(), Smooth().
  • dip.Histogram addition and subtraction operators, and indexing operators.
  • New functions dip.CumulativeHistogram(), dip.Smooth(), dip.Mean(), dip.Covariance(), dip.MarginalPercentile(), dip.MarginalMedian(), dip.Mode(), dip.PearsonCorrelation(), dip.Regression(), dip.MutualInformation(), dip.Entropy(), dip.IsodataThreshold(), dip.OtsuThreshold(), dip.MinimumErrorThreshold(), dip.GaussianMixtureModelThreshold(), dip.TriangleThreshold(), dip.BackgroundThreshold(), dip.KMeansClustering(), dip.MinimumVariancePartitioning(), dip.EqualizationLookupTable(), dip.MatchingLookupTable().
  • New function dip.PerObjectHistogram().

• Added dip.HistogramShow() and dip.Histogram.Show() to visualize histograms with proper axis scaling and tick labels.

• Added dip.LookupTable as a class, matching DIPlib’s dip::LookupTable.

• Added dip.ObjectMaximum() and dip.ObjectMinimum().

• Added dip.BlendBandlimitedMask().

• Added dip.Image.IsForged().

• Added the class dip.Graph, and the functions dip.RegionAdjacencyGraph() and dip.Relabel() with a graph as input.

Changed functionality

• The function dip.Histogram() is now called dip.Histogram_old(), to make space for the new dip.Histogram class. NOTE! This breaks backwards compatibility. To keep old code from working, do dip.Histogram = dip.Histogram_old at the top of the code, after import diplib as dip.

• The function dip.LookupTable() is now called dip.LookupTable_old(), to make space for the new dip.LookupTable class. NOTE! This breaks backwards compatibility. To keep old code from working, do dip.LookupTable = dip.LookupTable_old at the top of the code, after import diplib as dip.

• dip.HistogramMatching() now also accepts a dip.Histogram object as the second input argument, matching the DIPlib function.

• dip.Show() has a new parameter extent, uses proper warnings rather than just printing messages to the console, and no longer uses dip.ImageDisplay() for 1D images.

• The dip::StructuringElement class was exposed to Python as dip.SE (to make typing easier). It is now exposed as dip.StructuringElement, with dip.SE an alias for backwards compatibility (and for less typing).

• Added some constructor overloads and implicit conversions that make it much easier to work with pixel sizes.

(See also changes to DIPlib.)

Bug fixes

• Fixed memory corruption bug when releasing data from a different thread.

• Fixed name of last input argument to dip.GreyWeightedDistanceTransform, which is now mode.

• A tuple with numbers now implicitly converts to a dip.StructuringElement and a dip.Kernel.

(See also bugfixes to DIPlib.)

Changes to DIPviewer

New functionality

• Added possibility to build DIPviewer without GLFW or FreeGLUT. This must be explicitly enabled by setting DIP_BUILD_DIPVIEWER. This is useful when building DIPviewer for use in MATLAB, where Java provides the window functionality.

• Can now step dimensions by scrolling in the status bar when it is showing the operating point.

• Can now set the visualized dimensions programmatically from Python.

• Made window linking easier in Python.

• Added dip.Image.ShowSlice() as an alias for dip.viewer.Show().

• Better integration into the Python/IPython REPL (use %gui dip in IPython)

Bug fixes

• Fixed an issue that could cause a segmentation violation after all the DIPviewer windows were closed.

• Fixed an issue that prevented Visual Studio from building PyDIPviewer.

• Fixed range calculation for images with infinities.

Changes to DIPjavaio

None.