Preprocess Module

This module contains functions for image preprocessing before the main analysis. These functions include gaussian smoothing, histogram equalization, and background elimination.

The scout preprocess command can be used to apply these functions as well as convert to a Zarr array.

  • histogram : estimate image histogram

  • rescale : remove background below a threshold

  • denoise : remove noise using wavelet filtering

  • contrast : apply adaptive histrogram equalization (optional)

  • convert : convert image(s) to Zarr

scout.preprocess.clahe(image, kernel_size, clip_limit=0.01, nbins=256, nb_workers=None)

Apply CLAHE to each z-slice in image

Parameters

  • image (ndarray) – input image

  • kernel_size (int or list-like) – shape of the contextual regions

  • clip_limit (float, optional) – limit for number of clipping pixels

  • nbins (int, optional) – number of gray bins for histograms

  • nb_workers (int, optional) – number of workers to use. Default, cpu_count

Returns

equalized – output image

Return type

ndarray

scout.preprocess.denoise(image, sigma, wavelet='db1', nb_workers=None)

Denoise input image slice-by-slice with wavelet filtering

Parameters

  • image (ndarray) – Input 3D image

  • sigma (float) – Noise standard deviation

  • wavelet (str) – Wavelet to use in DWT. Default, ‘db1’.

  • nb_workers (int) – Number of parallel processes to use. Default, cpu_count()

Returns

output – Denoised 3D image

Return type

ndarray

scout.preprocess.denoise2d(image, sigma, wavelet='db1')

Denoise input image using wavelet filtering filtering

Parameters

  • image (ndarray) – Input 2D image

  • sigma (float) – Noise standard deviation

  • wavelet (str) – Wavelet to use in DWT. Default, ‘db1’.

Returns

output – Denoised 2D image

Return type

ndarray

scout.preprocess.gaussian_blur(image, sigma)

Smooth input image by gaussian blurring with sigma kernel.

This function does not normalize the image before applying the gaussian blurring.

Parameters

  • image (ndarray) – Input image

  • sigma (int, tuple) – Blurring amount for each axis

Returns

blurred – Smoothed image

Return type

ndarray

scout.preprocess.gaussian_blur_parallel(arr, sigma, output, chunks, overlap, nb_workers=None)

Smooths a chunked Zarr array with parallel processing.

Parameters

  • arr (Zarr Array) –

  • sigma (int, tuple) –

  • output (Zarr Array) –

  • chunks (tuple) –

  • overlap (int) –

  • nb_workers (int) –

scout.preprocess.remove_background(image, threshold, offset=None)

Threshold an image and use as a mask to remove the background. May improve compression

Parameters

  • image (ndarray) – input image

  • threshold (float) – intensity to threshold the input image

  • offset (float) – intensity to substract from the result

Returns

output – output image with background set to 0

Return type

ndarray