deepdisc.data_format.image_readers

Classes

`ImageReader`	Base class that will read images on the fly for the training/testing dataloaders
`DC2ImageReader`	An ImageReader for DC2 image files.
`HSCImageReader`	An ImageReader for HSC image files.
`RomanImageReader`	An ImageReader for Roman image files.

Module Contents

class ImageReader(norm='raw', *args, **kwargs)[source]

Bases: abc.ABC

Base class that will read images on the fly for the training/testing dataloaders

To implement an image reader for a new class, the derived class needs to have an __init__() function that calls super().__init__(*args, **kwargs) and a custom version of _read_image().

scaling[source]

scalekwargs[source]

abstract _read_image(key)[source]

Read the image. No-op implementation.

Parameters:: key (str or int) – The key indicating the image to read.
Returns:: im – The image.
Return type:: numpy array

__call__(image)[source]

Read the image and apply scaling.

Parameters:: image (str or numpy array) – The path indicating the image to read or image data in a numpy array with dimensions (band, h, w).
Returns:: im – The image.
Return type:: numpy array

raw()[source]

Apply raw image scaling (no scaling done).

Parameters:: im (numpy array) – The image.
Returns:: The image with pixels as float32.
Return type:: numpy array

lupton(bandlist=[2, 1, 0], stretch=0.5, Q=10, m=0)[source]

Apply Lupton scaling to the image and return the scaled image.

Parameters:

im (np array) – The image being scaled
bandlist (list[int]) – Which bands to use for lupton scaling (must be 3)
stretch (float) – lupton stretch parameter
Q (float) – lupton Q parameter
m (float) – lupton minimum parameter

Returns:

image – The 3-channel image after lupton scaling using astropy make_lupton_rgb

Return type:

numpy array

zscore(A=1.0, m=0.0)[source]

Apply z-score scaling to the image and return the scaled image.

Parameters:

im (np array) – The image being scaled
A (float) – A multiplicative scaling factor applied to each band
m (float) – A minimum pixel value. Defaults to 0.0

Returns:

image – The image after z-score scaling (subtract mean and divide by std deviation)

Return type:

numpy array

norm_dict[source]

classmethod add_scaling(name, func)[source]

Add a custom contrast scaling function

ex) def sqrt(image):

image[:,:,0] = np.sqrt(image[:,:,0]) image[:,:,1] = np.sqrt(image[:,:,1]) image[:,:,2] = np.sqrt(image[:,:,2]) return image

ImageReader.add_scaling(‘sqrt’,sqrt)

class DC2ImageReader(*args, **kwargs)[source]

Bases: ImageReader

An ImageReader for DC2 image files.

_read_image(filename)[source]

Read the image.

Parameters:: filename (str) – The filename indicating the image to read.
Returns:: im – The image.
Return type:: numpy array

class HSCImageReader(*args, **kwargs)[source]

Bases: ImageReader

An ImageReader for HSC image files.

_read_image(filenames)[source]

Read the image.

Parameters:: filenames (list) – A length 3 list of filenames for the I, R, and G images.
Returns:: im – The image.
Return type:: numpy array

class RomanImageReader(*args, **kwargs)[source]

Bases: ImageReader

An ImageReader for Roman image files.

_read_image(filename)[source]

Read the image.

Parameters:: filename (str) – The filename indicating the image to read.
Returns:: im – The image.
Return type:: numpy array