.. only:: html

    .. note::
        :class: sphx-glr-download-link-note

        Click :ref:`here <sphx_glr_download_examples_imaging_filter_images.py>`     to download the full example code
    .. rst-class:: sphx-glr-example-title

    .. _sphx_glr_examples_imaging_filter_images.py:


Image Filtering
===============
Learn how to filter images through a pipeline.


.. code-block:: default


    # Make sure we have the image plugin
    import numpy as np
    import imagepypelines as ip
    ip.require('image')









Construct our filtering blocks


.. code-block:: default


    # Frequency Filter
    @ip.blockify()
    def freq_filter(src, kernel):
        return src * kernel

    # Easy "frequency space" normalized lowpass filter
    @ip.blockify()
    def circular_pass_filter(shape, radius=0.1, type='low'):
        x, y = np.meshgrid(np.linspace(-1, 1, shape[0]), np.linspace(-1, 1, shape[1]))
        circle = np.sqrt(x*x + y*y)

        if type=='low':
            return np.where(circle > radius, 0, 1)

        elif type=='high':
            return np.where(circle > radius, 1, 0)

        else:
            raise ValueError("Filter 'type' must be either 'low' or 'high'")








Define our tasks and build the pipeline


.. code-block:: default


    tasks = {
            # set an entry point for images into the pipeline
            'images': ip.Input(),
            # get the image dimensions for later use
            'shape': (ip.image.Dimensions(order="WHC"), 'images'),
            # take the fourier transform
            'fft': (ip.image.ImageFFT(order="WHC"), 'images'),
            # produce 'frequency space' circular pass filters
            'circles': (circular_pass_filter, 'shape'),
            # add an extra dimension to the filters for broadcasting the arrays
            'unsqueezed': (ip.image.Unsqueeze(-1), 'circles'),
            # the convolution of your image and filter is element-wise
            # multiplication in frequency space!!!
            'filtered': (freq_filter, 'fft', 'unsqueezed'),
            # take the inverse fourier transform of your filtered images
            'ifft': (ip.image.ImageIFFT(order="WHC"), 'filtered'),
            # Scale the images, cast dtype, and view the filtered images
            # in sequence! Note: simple filtering in RGB results in artifacts!
            'safe': (ip.image.DisplaySafe(), 'ifft'),
            'null' : (ip.image.CompareView(pause_for=500), 'images','safe')
             }

    im_filt = ip.Pipeline(tasks)








Let's process some data!
------------------------

Let's grab some example data from the ImagePypelines standard set


.. code-block:: default

    images = [ip.image.panda(), ip.image.gecko(), ip.image.redhat()]
    # Number and view the images!
    processed = im_filt.process(images)



.. image:: /examples/imaging/images/sphx_glr_filter_images_001.png
    :alt: Image1, Image2
    :class: sphx-glr-single-img






.. rst-class:: sphx-glr-timing

   **Total running time of the script:** ( 0 minutes  2.312 seconds)


.. _sphx_glr_download_examples_imaging_filter_images.py:


.. only :: html

 .. container:: sphx-glr-footer
    :class: sphx-glr-footer-example



  .. container:: sphx-glr-download sphx-glr-download-python

     :download:`Download Python source code: filter_images.py <filter_images.py>`



  .. container:: sphx-glr-download sphx-glr-download-jupyter

     :download:`Download Jupyter notebook: filter_images.ipynb <filter_images.ipynb>`


.. only:: html

 .. rst-class:: sphx-glr-signature

    `Gallery generated by Sphinx-Gallery <https://sphinx-gallery.github.io>`_