virelay.image_processing

Contains some helper functions for processing images.

Functions

`add_border`	Up-samples the specified image, by making a border around the image.
`center_crop`	Crops the image evenly on all sides to the desired size.
`generate_heatmap_image_black_fire_red`	Generates a heatmap with a gray background, where red tones are used to visualize positive relevance values and blue tones are used to visualize negative relevances.
`generate_heatmap_image_black_green`	Generates a heatmap with a black background, where green tones are used to visualize positive relevance values and blue tones are used to visualize negative relevances.
`generate_heatmap_image_black_yellow`	Generates a heatmap with a black background, where yellow tones are used to visualize positive relevance values and blue tones are used to visualize negative relevances.
`generate_heatmap_image_gray_red`	Generates a heatmap with a gray background, where red tones are used to visualize positive relevance values and blue tones are used to visualize negative relevances.
`generate_heatmap_image_using_matplotlib`	Generates a heatmap from the specified attribution data using the color maps provided by matplotlib.
`render_heatmap`	Takes the raw attribution data and converts it so that the data can be visualized as a heatmap.
`render_superimposed_heatmap`	Renders the heatmap an superimposes it onto the specified image.

virelay.image_processing.add_border(image, new_width, new_height, method)[source]

Up-samples the specified image, by making a border around the image.

Parameters

image (numpy.ndarray) – The image that is to be up-sampled.
new_width (int) – The new width to which the image is to be up-sampled.
new_height (int) – The new height to which the image is to be up-sampled.
method (str) – The method that is to be used to create the border. Supported methods are ‘fill_zeros’ to fill up the border with zeros, ‘fill_ones’ to fill up the border with ones, ‘edge_repeat’ to repeat the values at the edge of the image, ‘mirror_edge’ to mirror the image at the edge, and ‘wrap_around’ to wrap the image around (e.g. the left edge is filled up with values from the right edge).

Raises

ValueError – If the specified method is not supported, then a ValueError is raised.

Returns

Returns the up-sampled image.

Return type

numpy.ndarray

virelay.image_processing.center_crop(image, new_width, new_height)[source]

Crops the image evenly on all sides to the desired size.

Parameters

image (numpy.ndarray) – The image that is to be down-sampled.
new_width (int) – The new width to which the image is to be down-sampled.
new_height (int) – The new height to which the image is to be down-sampled.

Returns

Returns the center-cropped image.

Return type

numpy.ndarray

virelay.image_processing.generate_heatmap_image_black_fire_red(attribution_data)[source]

Generates a heatmap with a gray background, where red tones are used to visualize positive relevance values and blue tones are used to visualize negative relevances.

Parameters: attribution_data (numpy.ndarray) – The raw attribution data for which the heatmap is to be rendered.
Returns: Returns the heatmap.
Return type: numpy.ndarray

virelay.image_processing.generate_heatmap_image_black_green(attribution_data)[source]

Generates a heatmap with a black background, where green tones are used to visualize positive relevance values and blue tones are used to visualize negative relevances.

Parameters: attribution_data (numpy.ndarray) – The raw attribution data for which the heatmap is to be rendered.
Returns: Returns the heatmap.
Return type: numpy.ndarray

virelay.image_processing.generate_heatmap_image_black_yellow(attribution_data)[source]

Generates a heatmap with a black background, where yellow tones are used to visualize positive relevance values and blue tones are used to visualize negative relevances.

Parameters: attribution_data (numpy.ndarray) – The raw attribution data for which the heatmap is to be rendered.
Returns: Returns the heatmap.
Return type: numpy.ndarray

virelay.image_processing.generate_heatmap_image_gray_red(attribution_data)[source]

Generates a heatmap with a gray background, where red tones are used to visualize positive relevance values and blue tones are used to visualize negative relevances.

Parameters: attribution_data (numpy.ndarray) – The raw attribution data for which the heatmap is to be rendered.
Returns: Returns the heatmap.
Return type: numpy.ndarray

virelay.image_processing.generate_heatmap_image_using_matplotlib(attribution_data, color_map_name)[source]

Generates a heatmap from the specified attribution data using the color maps provided by matplotlib.

Parameters

attribution_data (numpy.ndarray) – The raw attribution data for which the heatmap is to be rendered.
color_map_name (str) – The name of the color map that is used to produce the heatmap.

Returns

Returns the heatmap.

Return type

numpy.ndarray

virelay.image_processing.render_heatmap(attribution_data, color_map)[source]

Takes the raw attribution data and converts it so that the data can be visualized as a heatmap.

Parameters

attribution_data (numpy.ndarray) – The raw attribution data for which the heatmap is to be rendered.
color_map (str) – The name of color map that is to be used to render the heatmap.

Raises

ValueError – If the specified color map is unknown, then a ValueError is raised.

Returns

Returns the heatmap image.

Return type

numpy.ndarray

virelay.image_processing.render_superimposed_heatmap(attribution_data, superimpose, color_map)[source]

Renders the heatmap an superimposes it onto the specified image.

Parameters

attribution_data (numpy.ndarray) – The raw attribution data for which the heatmap is to be rendered.
superimpose (numpy.ndarray) – An image onto which the image is to be superimposed.
color_map (str) – The name of color map that is to be used to render the heatmap.

Raises

ValueError – If the specified color map is unknown, then a ValueError is raised.

Returns

Returns the heatmap image.

Return type

numpy.ndarray