pygorpho 0.2.0

Python bindings for gorpho

# pygorpho

Python bindings for [gorpho](https://github.com/patmjen/gorpho).

This is a Python library for fast 3D mathematical morphology using CUDA. Currently, the library provides: * Dilation and erosion for grayscale 3D images. * Support for flat or grayscale structuring elements. * A van Herk/Gil-Werman implementation for fast dilation/erosion with flat line segments in 3D. * Automatic block processing for 3D images which can’t fit in GPU memory.

Documentation can be found on [readthedocs](https://pygorpho.readthedocs.io)

## Installation First, make sure you have [CUDA Toolkit](https://developer.nvidia.com/cuda-toolkit) 9.2 or later installed. Also, you must install [NumPy](https://numpy.org/): ` pip install numpy ` Then, install with pip: ` pip install -i https://test.pypi.org/simple/ pygorpho `

## Installing from source Again, make sure you have [CUDA Toolkit](https://developer.nvidia.com/cuda-toolkit) 9.2 or later installed. Also, you need a compatible C++ compiler, which supports C++14. Then, following these instructions should allow you to build and install the package:

1. Clone the repo: git clone https://github.com/patmjen/pygorpho.git

2. Change directory: cd pygorpho

3. Install the required Python packages: pip install numpy scikit-build cmake ninja

4. Build and install: python setup.py install

That should be it! To test, navigate to another directory (e.g. cd ..), run python, and try to import pygorpho as pg.

Note: the reason you have to navigate to another directory before importing the package is because the build step makes a shared library. This will be installed with pygorpho, but if you stay in the folder where you cloned the git repo., import pygorpho will import from the local version, and not the installed version.