Quickstart Tutorial

See also

See our gallery of examples for specific applications of PySAP.

Basics

You can import the PySAP package as follows.

import pysap

Then you can print some basic information about PySAP,

print(pysap.info())

which should look something like this.

                 .|'''|       /.\      '||'''|,
                 ||          // \\      ||   ||
'||''|, '||  ||` `|'''|,    //...\\     ||...|'
 ||  ||  `|..||   .   ||   //     \\    ||
 ||..|'      ||   |...|' .//       \\. .||
 ||       ,  |'
.||        ''

Package version: 0.0.6

License: CeCILL-B

Authors:

Antoine Grigis <antoine.grigis@cea.fr>
Samuel Farrens <samuel.farrens@cea.fr>
Jean-Luc Starck <jl.stark@cea.fr>
Philippe Ciuciu <philippe.ciuciu@cea.fr>

Dependencies:

scipy          : >=1.5.4   - required | 1.7.3     installed
numpy          : >=1.19.5  - required | 1.22.1    installed
matplotlib     : >=3.3.4   - required | 3.5.1     installed
astropy        : >=4.1     - required | 5.0       installed
nibabel        : >=3.2.1   - required | 3.2.1     installed
pyqtgraph      : >=0.11.1  - required | 0.12.3    installed
progressbar2   : >=3.53.1  - required | ?         installed
modopt         : >=1.5.1   - required | 1.6.0     installed
scikit-learn   : >=0.24.1  - required | ?         installed
pywt           : >=1.1.1   - required | 1.2.0     installed
pysparse       : >=0.0.1   - required | 0.1.0     installed

Sample data

PySAP includes example data from various imaging domains that can be used for testing. Sample data sets can downloaded using the get_sample_data function.

Example

To download an image of galaxy NGC2997 you would run the following.

from pysap.data import get_sample_data
galaxy = get_sample_data('astro-ngc2997')

Image objects

The principal data type in PySAP are Image objects. These objects include a built-in show method.

Example

Running type(galaxy) with the object from the previous example should give you <class 'pysap.base.image.Image'> and thus you can directly visualise this object.

galaxy.show()

Image objects can easily be created from Numpy arrays.

Example

To create a PySAP Image object from a 2D numpy array you would do the following.

import numpy as np
from pysap import Image
my_array = np.arange(9).reshape((3, 3))
my_image = Image(data=my_array)

Transforms

You can display all of the available data transforms (i.e. wavelets and more) using the wavelist function.

Attention

We plan to provide more detailed information on each of the available transforms in the near future. The keys 'isap-3d' and 'isap-2d' from this dictionary correspond to Sparse2D, while 'pywt' corresponds to PyWavelets.

print(pysap.wavelist())

A given PySAP transform can be loaded using the load_transform function.

Example

To load the isotropic undecimated wavelet (or starlet) transform with 4 scales and symmetric padding you would do the following.

from pysap import load_transform
my_transform = load_transform('BsplineWaveletTransformATrousAlgorithm')(
    nb_scale=4,
    padding_mode='symmetric',
)

This will create a class instance of the desired transform with parent WaveletTransformBase, which has the methods analysis and synthesis that can be used to transform images into the corresponding domain and back again.

Tip

Transform objects also have access to the Image show method. Try this!

my_transform.show()

Example

Using the transform defined in the previous example and and the galaxy Image object from earlier you can transform this data into the wavelet domain as follows.

my_transform.data = galaxy
my_transform.analysis()