3 demo edge.py

import os
from athec import misc, edge, box

img_folder = os.path.join("image", "original")
resize_folder = os.path.join("image", "resize")
tf_folder = os.path.join("image", "transform")

imgname = "example1.jpg"
img = os.path.join(resize_folder, imgname)

'''
Perform Canny edge detection.
Return a 2-D array.
save_path (optional, default None): str. If provided, a visualization will be saved to this location.
otsu_ratio (optional, default 0.5): float. Canny edge detection needs input for lower and upper limits. The function will use Otsu method to find the lower and upper limits. otsu_ratio is the ratio between the lower limit and the upper limit.
thresholds (optional, default None): a tuple of two integers. If provided, the function will use these two numbers as the lower and upper limits. See https://docs.opencv.org/3.4/da/d22/tutorial_py_canny.html
gaussian_blur_kernel (optional, default None): a tuple of two integers. This argument specifies the width and height of the kernel of a Gausssian blur filter. Both numbers should be positive and odd. If this argument is provided, a Gausssian blur will be applied before edge detection. This step is to remove noise in the image. See https://docs.opencv.org/3.4/da/d22/tutorial_py_canny.html and https://docs.opencv.org/master/d4/d13/tutorial_py_filtering.html
'''

edges = edge.tf_edge_canny(img,
                           save_path = os.path.join(tf_folder, "edge canny", imgname),
                           thresholds = None, 
                           otsu_ratio = 0.5,
                           gaussian_blur_kernel = (5,5))

'''
Calculate visual complexity based on the edge map.
Return the percentage of image area occupied by edges and the average distance among all the pairs of edge points (divided by diagonal length).
n_random (optional, default None): int. If provided, the function will calculate the average distance among a random subset of edge points to speed up calculations.
'''

result = edge.attr_complexity_edge(edges,
                                   n_random = 1000)
misc.printd(result)

'''
This function can also take the file path of the edge image as the input.
'''
edge_path = os.path.join(tf_folder, "edge canny", imgname + '.png')
result = edge.attr_complexity_edge(edge_path)
misc.printd(result)

'''
Calculate visual complexity based on the relative size of a minimal bounding box that contains a certain percentage of edge points.
Return:
(1) The size of the box, divided by image size
(2) The coordinates of the box. The upper-left corner of the image is (0, 0).
save_path (optional, default None): str. If provided, a visualization will be saved to this location.
min_perentage (optional, default 0.9): float. The minimal percentage of edge points this bounding box should contain.
check_interval (optional, default 1): int. The interval the function will search by for the bounding box. Increasing this number can speed up calculations but results are less accurate.
'''
result = edge.attr_complexity_edge_box(edges,
                                      save_path = os.path.join(tf_folder, "bounding box edge canny", imgname),
                                      min_perentage = 0.9,
                                      check_interval = 1)

misc.printd(result)