nabu.processing.unsharp module

class nabu.processing.unsharp.UnsharpMask(shape, sigma, coeff, mode='reflect', method='gaussian')[source]

Bases: object

A helper class for unsharp masking.

Initialize a Unsharp mask. UnsharpedImage = (1 + coeff)*Image - coeff * ConvolutedImage

If method == “log”: UnsharpedImage = Image + coeff*ConvolutedImage

Parameters:

  • shape (tuple) – Shape of the image.

  • sigma (float) – Standard deviation of the Gaussian kernel

  • coeff (float) – Coefficient in the linear combination of unsharp mask

  • mode (str, optional) – Convolution mode. Default is “reflect”

  • method (str, optional) – Method of unsharp mask. Can be “gaussian” (default) or “log” (Laplacian of Gaussian), or “imagej”.

Notes

The computation is the following depending on the method:

  • For method=”gaussian”: output = (1 + coeff) * image - coeff * image_blurred

  • For method=”log”: output = image + coeff * image_blurred

  • For method=”imagej”: output = (image - coeff*image_blurred)/(1-coeff)

avail_methods = ['gaussian', 'log', 'imagej']
unsharp(image, output=None)[source]

Reference unsharp mask implementation.