Faxify

class augraphy.augmentations.faxify.Faxify(scale_range=(1.0, 1.25), monochrome=-1, monochrome_method='random', monochrome_arguments={}, halftone=-1, invert=1, half_kernel_size=(1, 1), angle=(0, 360), sigma=(1, 3), numba_jit=1, p=1)

Bases: Augmentation

Emulates faxify effect in the image.

Parameters:

• scale_range (tuple, optional) – Pair of floats determining the range from which to divide the resolution by.

• monochrome (int, optional) – Flag to enable monochrome effect.

• monochrome_method (string, optional) – Monochrome thresholding method.

• monochrome_arguments (dict, optional) – A dictionary contains argument to monochrome thresholding method.

• halftone (int, optional) – Flag to enable halftone effect.

• invert (int, optional) – Flag to invert grayscale value in halftone effect.

• half_kernel_size (tuple, optional) – Pair of ints to determine half size of gaussian kernel for halftone effect.

• angle (tuple, optional) – Pair of ints to determine angle of halftone effect.

• sigma (tuple, optional) – Pair of ints to determine sigma value of gaussian kernel in halftone effect.

• numba_jit (int, optional) – The flag to enable numba jit to speed up the processing in the augmentation.

• p (float, optional) – The probability that this Augmentation will be applied.

static apply_halftone(image_halftone, gaussian_kernel, rotated, ysize, xsize, kernel_size)

Run loops and apply halftone effect in the input image.

Parameters:

• image_halftone (numpy.array (numpy.uint8)) – The image with halftone effect.

• gaussian_kernel (numpy.array (numpy.uint8)) – Gaussian kernel to generate the halftone effect.

• rotated (numpy.array (numpy.uint8)) – The rotated input image

• ysize (int) – Row numbers of the input image.

• xsize (int) – Column numbers of the input image.

• kernel_size (int) – Kernel size to generate the halftone effect.

complement_rgb_to_gray(img, invert=1, gray_level=255, max_value=255)

Convert RGB/BGR image to single channel grayscale image.

Parameters:

• img (numpy.array (numpy.uint8)) – The image to apply the function.

• invert (int) – Flag to invert the generated grayscale value.

• gray_level (int) – The selected gray value.

• max_value (int) – Maximum value of gray value.

cv_rotate(image, angle)

Rotate image based on the input angle.

Parameters:

• image (numpy.array (numpy.uint8)) – The image to apply the function.

• angle (int) – The angle of the rotation.

downscale(image)

Downscale image based on the user input scale value.

Parameters:

image (numpy.array (numpy.uint8)) – The image to apply the function.

generate_halftone(image, half_kernel_size=2, angle=45, sigma=2)

Generate halftone effect in the input image.

Parameters:

• image (numpy.array (numpy.uint8)) – The image to apply the function.

• half_kernel_size (int) – Half value of kernel size to generate halftone effect.

• angle (int) – The angle of the halftone effect.

• sigma (int) – Sigma value of gaussian kernel for halftone effect.

Overview

The Faxify augmentation emulates the artifacts created by faxing the document.

Initially, a clean image with single line of text is created.

Code example:

# import libraries
import cv2
import numpy as np
from augraphy import *


# create a clean image with single line of text
image = np.full((500, 1500,3), 250, dtype="uint8")
cv2.putText(
    image,
    "Lorem ipsum dolor sit amet, consectetur adipiscing elit",
    (80, 250),
    cv2.FONT_HERSHEY_SIMPLEX,
    1.5,
    0,
    3,
)

cv2.imshow("Input image", image)

Clean image:

Example 1

In this example, a Faxify augmentation instance is initialized and it is set to opencv thresholding method (“cv2.threshold”) for monochrome effect. Halftone effect is enable (1) and halftone half kernel size is set to 2 (2,2). Halftone angle is set randomly between 0 to 360 degree (0, 360) and the sigma value of Gaussian blur is to random value between 1 and 3 (1,3).

Code example:

faxify = Faxify(scale_range = (1,2),
                monochrome = 1,
                monochrome_method = "cv2.threshold",
                monochrome_arguments = {"thresh":128, "maxval":128, "type":cv2.THRESH_BINARY},
                halftone = 1,
                invert = 1,
                half_kernel_size = (2,2),
                angle = (0, 360),
                sigma = (1,3))

img_faxify = faxify(image)
cv2.imshow("faxify", img_faxify)

Augmented image: