Release Notes: This release adds a color conversion module (mahotas.color) as well as a basic SLIC super-pixels implementation and a few extra label handling functions: relabel and remove_regions. There are bugfixes for distance function and median_filter.
Release Notes: This version adds several new morphological functions: subm, cdilate, and tophat transforms. It adds border handling to functions where it was missing, such as border, euler, and bwperim. It includes a few fixes for older compilers.
Release Notes: There has been a namespace reorganisation. For simple uses, it shouldn't matter, but you might find that some functions have moved submodules. An important bug in dilate() has been fixed. Also new are a few morphological functions: open(), close(), regmax(), and regmin().
Release Notes: This release fixes compilation on Mac OS X. It adds the convolve1d function and fixes minor issues.
Release Notes: This version contains two major new features: SURF local features and an implementation of n-dimensional convolution. It also contains a few minor new features: it releases the GIL in morphological functions, adds a just_filter option in edge.sobel(), and has a new submodule mahotas.labeled to deal with labeled images.
Release Notes: Improved local binary patterns (a faster and better interface). Much faster erode() (10x faster). Faster dilate() (2x faster). TAS for 3D images. Haralick for 3D images.
Release Notes: This release implements mahotas.thin() in C++ for speed. It adds np.uint64 support to fullhistogram(). It will drop GIL when possible. It adds an imresize() wrapper around ndimage.zoom.
Release Notes: The dependency on freeimage is now optional. bbox() has been extended to N>2 dimensions. There are documentation fixes.
Release Notes: Major Changes: distance transform, bwperim(), a freeimage interface, and zernike moment computation. Minor changes: namespace reorganization, many bugfixes, and speed improvements in many functions.
Release Notes: The major new features are support for 64-bit images in many algorithms, faster watershed, hit & miss, thinning, threshold adjacency statistics, and polygon filling routines. Small fixes were made all around.