2.1.2.1.2. sf_tools.image.distort module

IMAGE DISTORTION ROUTINES

This module contains methods for playing around with image properties.

Author:Samuel Farrens <samuel.farrens@gmail.com>
Version:1.1
Date:20/10/2017
sf_tools.image.distort.downsample(image, factor)[source]

Downsample

This method downsamples (decimates) an image.

Parameters:
  • image (np.ndarray) – Input image array
  • factor (int) – Downsampling factor
Returns:
Return type:

np.ndarray downsampled image array
Raises:

ValueError – For invalid downsampling factor
sf_tools.image.distort.resize_even_image(image)[source]

Resize even image

This method returns an image with odd dimensions.

Parameters:image (np.ndarray) – Input image array
Returns:
Return type:np.ndarray resized image array
sf_tools.image.distort.roll_2d(data, roll_rad=(1, 1))[source]

Roll in 2D

This method rolls an array in 2 dimensions.

Parameters:
  • data (np.ndarray) – Input 2D data array
  • roll_rad (tuple) – Roll radius in each dimension
Returns:
Return type:

np.ndarray rolled array
sf_tools.image.distort.rot_and_roll(data)[source]

Rotate and roll

This method rotates (by 180 deg) and rolls a 2D array.

Parameters:data (np.ndarray) – Input 2D data array
Returns:
Return type:np.ndarray rotated and rolled array
sf_tools.image.distort.gen_mask(kernel_shape, image_shape)[source]

Generate mask

This method generates an image mask.

Parameters:
  • kernel_shape (tuple) – Shape of kernel
  • image_shape (tuple) – Shape of image
Returns:
Return type:

np.ndarray boolean mask
sf_tools.image.distort.roll_sequence(data_shape)[source]

Roll sequence

This method generates the roll sequence for a 2D array.

Parameters:data_shape (tuple) – Shape of data
Returns:
Return type:list of roll radii
sf_tools.image.distort.kernel_pattern(kernel_shape, mask)[source]

Kernel pattern

This method generates the kernel pattern. Rather than padding the kernel with zeroes to match the image size one simply extracts the series of repitions of the base kernel patterns.

Parameters:
  • kernel_shape (tuple) – Shape of kernel
  • mask (np.ndarray) – Boolean mask
Returns:
Return type:

np.ndarray kernel pattern
sf_tools.image.distort.rearrange_kernel(kernel, data_shape=None)[source]

Rearrange kernel

This method rearanges the input kernel elements for vector multiplication. The input kernel is padded with zeroes to match the image size.

Parameters:
  • kernel (np.ndarray) – Input kernel array
  • data_shape (tuple) – Shape of the data
Returns:
Return type:

np.ndarray rearanged matrix of kernel elements