imagediffer.core package

This package contains tools for loading and saving images, various color manipulations and comparing images.

imagediffer.core.differ module

Differ module contains functions for comparing images and creating diff images

imagediffer.core.differ.calculate_mse(image1, image2)

Calculate mean squared error for given images. The higher the value of MSE is, the more different the images are.

Parameters:
  • image1 – Numpy image array
  • image2 – Numpy image array
Returns:

Float number representing mean squared error
imagediffer.core.differ.calculate_ssim(image1, image2)

Calculate structural similarity index for given images. If the value is 1.0 the images are same. The lower the value is, the more different the images are.

Parameters:
  • image1 – Numpy image array
  • image2 – Numpy image array
Returns:

Float number from -1.0 to 1.0 representing SSIM
imagediffer.core.differ.chebyshev_distance(image1, image2)

Calculate chebyshev distance for each pixel in image1 and image2. Images must have same dimension.

Parameters:
  • image1 – Numpy image array
  • image2 – Numpy image array
Returns:

Array of chebyshev distances
imagediffer.core.differ.diff(image1, image2, compare_colors_method, tolerance=0, diff_color=(1.0, 0, 1.0, 1.0))

Create a diff image of image1 and image2. Pixels are considered different if the distance of colors is greater than tolerance using given compare_colors_method. The result image is created by blending image1 and image2 together and replacing different pixels with the diff_color.

Parameters:
  • image1 – Numpy image array
  • image2 – Numpy image array
  • compare_colors_method – Method that takes two numpy image arrays and return array of color distances in range from 0.0 to 1.0. You can use euclidean_distance, chebyshev_distance or implement your own method.
  • tolerance – Defines the color distance that is acceptable and colors are considered the same
  • diff_color – RGBA color that should be used for different pixels
Returns:

Tuple containing:
  • diff_image Numpy image array
  • diff_pctg Percentage of pixels where the color distance exceeded the acceptable tolerance
imagediffer.core.differ.euclidean_distance(image1, image2)

Calculate euclidean distance for each pixel in image1 and image2. Images must have same dimension.

Parameters:
  • image1 – Numpy image array
  • image2 – Numpy image array
Returns:

Array of euclidean distances

imagediffer.core.loader module

Loader module contains functions for loading and saving images.

imagediffer.core.loader.load_image_from_file(file)

Load image as a numpy array from given file.

Parameters:file – Path to the image file
Returns:Numpy image array with color and alpha channels values from 0.0 to 1.0
imagediffer.core.loader.load_image_from_url(url)

Load image as a numpy array from given URL.

Parameters:url – Image URL
Returns:Numpy image array with color and alpha channels values from 0.0 to 1.0
imagediffer.core.loader.save_image(image, path)

Save given image to the PNG file. If the file extension is not part of the path or is missing, it will be autocompleted or replaced.

Parameters:
  • image – Numpy image array
  • path – Path to the image file.

imagediffer.core.utils module

Utils module contains various functions for manipulating images.

imagediffer.core.utils.extract_colors(image)

Extract individual color channels and alpha channel from numpy image array.

Parameters:image – Numpy image array
Returns:Array of width × height numpy arrays for each color and alpha channels
imagediffer.core.utils.norm_color(image)

Normalize colors to be in range from 0.0 to 1.0 instead of from 0 to 255.

Parameters:image – Numpy image array with colors from 0 to 255
Returns:Numpy image array with colors normalized to floats
imagediffer.core.utils.to_blue(source)

Convert source image to image using blue channel for all color channels.

Parameters:source – Numpy image array
Returns:Numpy image array
imagediffer.core.utils.to_grayscale(source)

Convert image to grayscale using PAL/NTSC conversion.

Parameters:source – Numpy image array
Returns:Numpy image array
imagediffer.core.utils.to_green(source)

Convert source image to image using green channel for all color channels.

Parameters:source – Numpy image array
Returns:Numpy image array
imagediffer.core.utils.to_red(source)

Convert source image to image using red channel for all color channels.

Parameters:source – Numpy image array
Returns:Numpy image array