filtration package

Submodules

filtration.filter module

class filtration.filter.Filter

Bases: abc.ABC

A abstract base class for all Filters

abstract filter(region) → bool: details of the behavior of the Filter

class filtration.filter.FilterBlackAndWhite(filter_threshold=0.5, binarization_threshold=0.85, rgb_weights=[0.2989, 0.587, 0.114])

Bases: filtration.filter.Filter

Filters out regions with too much whitespace

convert_rgb_to_greyscale(region)

convert_rgb_to_greyscale Convert an RGB region of image to greyscale

Parameters: region (np.ndarray) – numpy array representing the region, region consists of RGB values
Returns: a numpy array representing the region, region is in greyscale
Return type: np.ndarray

filter(region) → bool

filter Perform filtration to a region

Parameters: region (np.ndarray) – numpy array representing the region
Returns: True if the average of the binary region is less than the filter threshold, else False
Return type: bool

class filtration.filter.FilterFocusMeasure(threshold=65.0)

Bases: filtration.filter.Filter

Filters out regions that are sufficiently out-of-focus

filter(region) → bool

filter Perform filtration to a region by determining whether it is focused or blurry

Parameters: region (np.ndarray) – numpy array representing the region
Returns: True if the focus measure is greater than the supplied threshold (image is not considered blurry), else False (image is considered blurry)
Return type: bool

variance_of_laplacian(region) → float

variance_of_laplacian Computes the Laplacian of the region

Parameters: region (np.ndarray) – numpy array representing the region
Returns: The focus measure which is the variance of the Laplacian
Return type: float

class filtration.filter.FilterHSV(threshold: numbers.Number = 100)

Bases: filtration.filter.Filter

Filters out regions according to average hue

convert_rgb_to_hsv(region) → numpy.ndarray

convert_rgb_to_hsv Converts RGB region of image to HSV

Parameters: region (np.ndarray) – numpy array representing the region, region consists of RGB values
Returns: a numpy array representing the region, region consists of HSV values
Return type: np.ndarray

filter(region) → bool

filter Perform filtration to a region

Parameters: region (np.ndarray) – numpy array representing the region
Returns: True if the mean of the hues in the hsv region is greather than the threshold, else False
Return type: bool

class filtration.filter.FilterHistolab(tissue_percentage: float = 0.8)

Bases: filtration.filter.Filter

Applies basic tissue filtration from the Histolab PyPi package

filter(region) → bool

filter filters regions based on the histolab package’s Tile.has_enough_tissue()

Parameters: region (np.ndarray) – the region to which filtration will be applied

class filtration.filter.FilterRandom(p: numbers.Number = 0.5)

Bases: filtration.filter.Filter

Decides whether a region passes randomly

filter(region) → bool

filter filters based on a random number being greater than the set threshold

Parameters: region (np.ndarray) – unused region of an image
Returns: Returns a boolean depending on whether RNG picks a number greater than the set threshold
Return type: bool

filtration.filter_manager module

class filtration.filter_manager.FilterManager(filters: Union[filtration.filter.Filter, str, list])

Bases: object

add_filter(filter: Union[filtration.filter.Filter, str])

add_filter Appends the input filter to the list of filters in state

Parameters

filter (Union[Filter, str]) – Either a Filter subclass or a str representing the filter’s type

Raises

Exception – This filter does not exist.
TypeError – This is not a filter.

filter(region: numpy.ndarray) → bool

filter Filters the region based on the input filters

Parameters: region (np.ndarray) – A region of the image to be filtered.
Returns: True if the region passes all of the filters, and false if not.
Return type: bool