Filtering Tools

Astroimtools provides functions for generating circular and elliptical (or annulus versions of each) footprint arrays, which can be used to perform specialized image filtering. The functions are:

A pixel is considered to be entirely in or out of a particular footprint depending on whether its center is in or out of the footprint. The size of the output array is the minimal bounding box for the footprint.

An example of use case of these footprints is median filtering an image using a circular (or elliptical) ring filter (see below).

Getting Started

Let’s generate footprint arrays using each function.

Here’s a circular footprint with radius of 2 pixels:

>>> from astroimtools import circular_footprint
>>> circular_footprint(2)
array([[0, 0, 1, 0, 0],
       [0, 1, 1, 1, 0],
       [1, 1, 1, 1, 1],
       [0, 1, 1, 1, 0],
       [0, 0, 1, 0, 0]])

Here’s a circular annulus footprint with an inner radius of 1 pixel and an outer radius of 2 pixels:

>>> from astroimtools import circular_annulus_footprint
>>> circular_annulus_footprint(1, 2)
array([[0, 0, 1, 0, 0],
       [0, 1, 0, 1, 0],
       [1, 0, 0, 0, 1],
       [0, 1, 0, 1, 0],
       [0, 0, 1, 0, 0]])

Here’s an elliptical footprint with a semimajor axis of 4 and a semiminor axis of 2 where the semimajor axis is rotated 90 degrees counterclockwise from the positive x axis (note that the data values are printed with y=0 at the top such that the array appears vertically flipped):

>>> import numpy as np
>>> from astroimtools import elliptical_footprint
>>> elliptical_footprint(4, 2, theta=np.pi/2.)
array([[0, 0, 1, 0, 0],
       [0, 1, 1, 1, 0],
       [0, 1, 1, 1, 0],
       [0, 1, 1, 1, 0],
       [1, 1, 1, 1, 1],
       [0, 1, 1, 1, 0],
       [0, 1, 1, 1, 0],
       [0, 1, 1, 1, 0],
       [0, 0, 1, 0, 0]])

Here’s an elliptical annulus footprint with an inner semimajor axis of 2, an outer semimajor axis of 4, and an inner semiminor axis of 1 where the semimajor axis is rotated 90 degrees counterclockwise from the positive x axis:

>>> import numpy as np
>>> from astroimtools import elliptical_annulus_footprint
>>> elliptical_annulus_footprint(2, 4, 1, theta=np.pi/2.)
array([[0, 0, 1, 0, 0],
       [0, 1, 1, 1, 0],
       [0, 1, 0, 1, 0],
       [0, 1, 0, 1, 0],
       [1, 0, 0, 0, 1],
       [0, 1, 0, 1, 0],
       [0, 1, 0, 1, 0],
       [0, 1, 1, 1, 0],
       [0, 0, 1, 0, 0]])

Ring Median Filter

The circular_annulus_footprint() function can be used to generate a ring-shaped footprint to implement a ring median filter. The effect of the ring filter is to remove objects from an image which have a size less than the inner radius and replace them with an estimate of the local background value (determined by the median image values within the footprint).

First, let’s create a ring filter with an inner radius of 10 pixels and and outer radius of 12 pixels. This will filter objects with a size of 10 pixels or less:

>>> from astroimtools import circular_annulus_footprint
>>> fp = circular_annulus_footprint(10, 12)

We can now use this footprint with the median_filter() function from Scipy:

>>> from scipy.ndimage import median_filter
>>> result = median_filter(data, footprint=fp)

Several additional filters are available in scipy.ndimage that accept footprint inputs.

Reference/API

Image filtering utilities.

Functions

circular_footprint(radius[, dtype])

Create a circular footprint.

circular_annulus_footprint(radius_inner, ...)

Create a circular annulus footprint.

elliptical_footprint(a, b[, theta, dtype])

Create an elliptical footprint.

elliptical_annulus_footprint(a_inner, ...[, ...])

Create an elliptical annulus footprint.