This repository is for the analysis of PIV images using adaptive PIV image analysis approaches

It is assumed that you have python set up on your machine already. If that is not the case, then look here to get started. Before installing AdaptivePIV, you will require a Python 3.7 environment with numpy installed. During installation a number of other packages will be installed - see below.

I would strongly recommend to use a virtual environment for this:

python -m venv env_name
env_name\Scripts\activate.bat

Or

conda create -n env_name python=3.7
conda activate env_name

Where in both cases env_name represents the name of the virtual environment. For more information see Python virtual environments or Conda virtual environments.

Numpy is trivially installed using

pip install numpy

or

conda install numpy

NOTE While numpy is required to install the package, a number of other packages are required to run the code. These will be automatically installed when installing the PIV package. The installed packages are:

• scipy
• h5py
• pillow
• sklearn
• scikit-image
• matplotlib
• bottleneck

Checkout the repository using

git clone https://github.com/MattEdwards94/AdaptivePIV.git

There are 2 options to install AdaptivePIV:

To install a static version, the distributable must first be built.

python setup.py sdist

This will build a tar.gz file under the sub-directory dist\ Installing the package and it's dependencies is then achieved via:

pip install --upgrade dist\PIV-x.x.tar.gz

Where x.x is the version number being installed.

Alternatively, the package can be installed in developer mode. In doing so, changes to the source code are automatically reflected when running the code without having to rebuild (Note that you will likely have to restart the Python interpreter)

pip install -e .

The python package Cython is used in this repository to accelerate certain aspects of the cross-correlation analysis. Files with the ending .pyx are cythonized into an equivalent C code which is built when the package is installed.
This C code is shipped with the repository and therefore the .pyx should not need building again.
If you would like to modify these Cython files then you must first install Cython:

pip install cython  / conda install cython

Following this, the package will need rebuilding with an additional flag:

python setup.py sdist --cyth True

For some unknown reason, installing the 3rd party package pillow from the conda repository results in tiff files being unable to be opened. This is resolved by installing the package using pip. A similar issue is encountered when installing jupyter using conda - I suspect that there is a dependency issue somewhere along the chain.

Included in this repository are two example PIV image pairs taken from an experiment of the flow over a backwards facing step. The function experimental_example(im_number=1, settings=None) in the file example.py will analyse one of these pairs. Options for im_number are 1(default) and 20.
If no settings are passed to this function, the following pre-configured defaults will be adopted:

WidimSettings(init_WS=97,
              final_WS=33,
              WOR=0.5,
              n_iter_main=3,
              n_iter_ref=1,
              vec_val='NMT',
              interp='struc_cub')

There are two ways to see this function in action. Calling python example.py will run both the experimental example and the synthetic example (below) one after another. Alternatively, to just run one, launch a python session from within the AdaptivePIV directory and import example, followed by running the desired function example.experimental_example() or example.synthetic_example().

If you would like to play about with the various settings, run help(PIV.analysis.WidimSettings.__init__) (with PIV imported) to see what each value can accept.

The synthetic_example(settings=None) shows how artificial images can be created and further analysed. By creating these artificial images, novel algorithms can be tested and validated against the known underlying displacement field.
The displacement field used in this example is of a non-physical contra rotating vortex array.
As above, this can either be seen by running the entire file python example.py or by importing example as a module and running example.synthetic_example()

If you have any issues with the above steps, or questions about the PIV analysis algorithm, please feel free to email me at [email protected].