IQM_Vis.transforms package

Submodules

IQM_Vis.transforms.additive_noise module

noise transformations

class IQM_Vis.transforms.additive_noise.Gaussian_noise(acceptable_percent=0.9, max_iter=50, reject_low_noise=False, **kwargs)

Bases: IQM_Vis.transforms.additive_noise._base_noise

Add Gaussian noise to image. Image is clipped to the range (0, 1)

Parameters

• image (np.array) – image to have Gaussian noise added

• std (float) – Standard deviation of the Gaussian noise. (Defaults to 0)

Returns

adjusted image

Return type

image (np.array)

class IQM_Vis.transforms.additive_noise.noise_hypersphere(acceptable_percent=0.9, max_iter=50, reject_low_noise=False, **kwargs)

Bases: IQM_Vis.transforms.additive_noise._base_noise

Add random noise on a hypersphere with radius epsilon. Image is clipped to the range (0, 1)

Parameters

• image (np.array) – image to have noise added

• std (float) – epsilon of the hypersphere radius of noise. (Defaults to 0)

Returns

adjusted image

Return type

image (np.array)

IQM_Vis.transforms.additive_noise.salt_and_pepper_noise(image, prob=0)

Add salt and pepper noise to image

Parameters

• image (np.array) – image to be compressed

• prob – Probability of the noise (Defaults to 0).

Returns

image with salt and pepper noise

Return type

image (np.array)

IQM_Vis.transforms.affine module

affine (geometric) transformations

IQM_Vis.transforms.affine.rotation(image, angle=0)

Rotate an image around its centre. Uses skimage.transform.rotate. Areas that are rotated beyond the image are filled in with black pixel values

Parameters

• image (np.array) – image to be rotated

• angle (float) – amount of rotation in degrees (Defaults to 0)

Returns

rotated image

Return type

image (np.array)

IQM_Vis.transforms.affine.x_shift(image, x_shift=0)

Translate image horizontally

Parameters

• image (np.array) – image to be shifted

• x_shift (float) – shift proportion of the image in the range (-1, 1). (Defaults to 0)

Returns

shifted image

Return type

image (np.array)

IQM_Vis.transforms.affine.y_shift(image, y_shift=0)

Translate image vertically

Parameters

• image (np.array) – image to be shifted

• y_shift (float) – shift proportion of the image in the range (-1, 1). (Defaults to 0)

Returns

shifted image

Return type

image (np.array)

IQM_Vis.transforms.affine.zoom_image(image, scale_factor=1)

digital zoom of image

Parameters

• image (np.array) – image to be zoomed

• scale_factor (float) – the percentage to zoom in by (for square zoom only). 0.5 = 2x zoom out 1 = normal size 2 = 2x zoom in (Defaults to 1)

Returns

zoomed image

Return type

image (np.array)

IQM_Vis.transforms.effects module

Sample image transformations to get the user started with

IQM_Vis.transforms.effects.binary_threshold(image, threshold=100)

conver image to binary at a given threshold

Parameters

• image (np.array) – image to be thresholded

• threshold (int) – threshold value in the range (1, 255) (Defaults to 100)

Returns

thresholded image (float32 in range 0, 1)

Return type

image (np.array)

IQM_Vis.transforms.effects.blur(image, kernel_size=1)

Gaussian Blur on an image

Parameters

• image (np.array) – image to be rotated

• kernel_size (int) – size of the convolutional kernel, will be converted to an odd number. 1 is no blur. (Defaults to 1)

Returns

rotated image

Return type

image (np.array)

IQM_Vis.transforms.effects.brightness(image, value=0)

Adjust image brightness. Image is clipped to the range (0, 1)

Parameters

• image (np.array) – image to have its brightness adjusted

• value (float) – value to adjust all pixels by in the range (-1, 1). (Defaults to 0)

Returns

adjusted image

Return type

image (np.array)

IQM_Vis.transforms.effects.brightness_hsv(image, b=0)

adjust the brightness of an image

Parameters

• image (np.array) – image to have brightness adjusted

• b (float) – amount of brightness to adjust by. (Defaults to 0).

Returns

image with brightness adjusted

Return type

image (np.array)

IQM_Vis.transforms.effects.contrast(image, contrast=1.0)

adjust the contrast of an image

Parameters

• image (np.array) – image to have contrast adjusted

• contrast (float) – amount of contrast to adjust by. Values less that 1 will decrease contrast, Values higher than 1 will increase the contrast. (Defaults to 1.0).

Returns

image with contrast adjusted

Return type

image (np.array)

IQM_Vis.transforms.effects.hue(image, h=0)

adjust the hue of an image

Parameters

• image (np.array) – image to have hue adjusted

• h (float) – amount of hue to adjust by. (Defaults to 0).

Returns

image with hue adjusted

Return type

image (np.array)

IQM_Vis.transforms.effects.jpeg_compression(image, compression=101)

encode image using jpeg then decode

Parameters

• image (np.array) – image to be compressed

• compression (int) – amount of jpeg compression, higher is better quality. 101 returns identity image. (Defaults to 101)

Returns

jpeg compressed image

Return type

image (np.array)

IQM_Vis.transforms.effects.saturation(image, sat=0)

adjust the saturation of an image

Parameters

• image (np.array) – image to have saturation adjusted

• sat (float) – amount of saturation to adjust by. (Defaults to 0).

Returns

image with saturation adjusted

Return type

image (np.array)

Module contents

IQM_Vis.transforms.get_all_transforms()

Get all the available transformation/distortions alongside their recommended parameter ranges

Returns

names, min, max, init_value and function for each transform

Return type

• all (dict)