EdgeDetectionImplementation/project1.py at main · Dominicpham03/EdgeDetectionImplementation · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
import cv2
import numpy
import math

def load_image(file_name):
    image = cv2.imread(file_name, 0)
    image = cv2.resize(image, (800, 1200))
    return image

def display_image(image):
    window_name = 'Display window'
    cv2.namedWindow(window_name, cv2.WINDOW_NORMAL)
    cv2.resizeWindow(window_name, 800, 1200)
    cv2.imshow(window_name, image)
    cv2.waitKey(0)
    cv2.DestroyAllWindows()

def generate_gaussian(sigma, filter_w, filter_h):
    mu = 0
    pi = math.pi
    two_sigma_squared = 2*(sigma**2)

    if(filter_w == 1 or filter_h == 1):
        filter_len = 0
        if filter_w >= filter_h:
            filter_len = filter_w
        else:
            filter_len = filter_h

        center = filter_len // 2
        result = numpy.zeros(filter_len)
        gaussian_value = 1 / (math.sqrt(2 * pi) * sigma)

        for i in range(filter_len):
            dist = (i - center)
            result[i] = gaussian_value * math.exp(-((dist-mu)**2) / two_sigma_squared)

        return result

    else:
        center_w = filter_w // 2
        center_h = filter_h // 2
        result = numpy.zeros((filter_h, filter_w))
        gaussian_value = 1 / (two_sigma_squared*pi)

        for i in range(filter_h):
            for j in range(filter_w):
                dist_x = (j - center_w)**2
                dist_y = (i - center_h)**2
                result[i][j] = gaussian_value * math.exp(-((dist_x + dist_y)) / two_sigma_squared)

        result /= numpy.sum(result)

        return result

def apply_filter(image, filter, pad_pixels, pad_value):
    image_h, image_w = image.shape
    output = numpy.zeros_like(image)

    if pad_value == 0:
        mode = 'constant'
    else:
        mode = 'edge'

    if filter.ndim == 1:
        pad_image = numpy.pad(image, ((0,0), (pad_pixels, pad_pixels)), mode)
        filter_len = filter.shape[0]

        for i in range(image_h):
            for j in range(image_w-filter_len + 1):
                area = pad_image[i, j:j + filter_len]
                output[i,j + pad_pixels] = numpy.sum(area * filter)

    else:
        filter_h, filter_w = filter.shape

        pad_image = numpy.pad(image, ((pad_pixels, pad_pixels), (pad_pixels, pad_pixels)), mode)

        for i in range(image_h):
            for j in range(image_w):
                area = pad_image[i:i + filter_h, j:j + filter_w]
                if area.shape == filter.shape:
                    output[i,j] = numpy.sum(area * filter)

    return output

def median_filtering(image, filter_w, filter_h):
    if filter_w % 2 == 0:
        filter_w += 1
    if filter_h % 2 == 0:
        filter_h += 1

    image_h, image_w = image.shape
    pad_w = filter_w // 2
    pad_h = filter_h // 2

    pad_image = numpy.pad(image, ((pad_h, pad_h), (pad_w, pad_w)), mode = 'edge')
    output = numpy.zeros_like(image)

    for j in range(image_h):
        for i in range(image_w):
            area = pad_image[j:j + filter_h, i:i + filter_w]
            output[j,i] = numpy.median(area)

    return output

def hist_eq(image):
    image_h, image_w = image.shape
    total_pixels = image_h * image_w
    hist_arr = numpy.zeros(256)
    output = numpy.zeros_like(image)

    for i in range(image_h):
        for j in range(image_w):
            hist_arr[image[i,j]] += 1

    for i in range(1, 256):
        hist_arr[i] += hist_arr[i - 1]

    for i in range(image_h):
        for j in range(image_w):
            output[i,j] = numpy.uint8(hist_arr[image[i,j]] * 255 / total_pixels)

    return output

def rotate(image, theta):
    image_h = image.shape[0]
    image_w = image.shape[1]
    center_h = image_h // 2
    center_w = image_w // 2

    rotated_w = round(abs(image_h * math.sin(theta)) + abs(image_w * math.cos(theta)))
    rotated_h = round(abs(image_h * math.cos(theta)) + abs(image_w * math.sin(theta)))

    rotated_image = numpy.zeros((rotated_h, rotated_w), dtype=numpy.uint8)

    new_center_h = rotated_h // 2
    new_center_w = rotated_w // 2

    for i in range(rotated_h):
        for j in range(rotated_w):
            original_x = int((j - new_center_w) * math.cos(-theta) - (i - new_center_h) * math.sin(-theta) + center_w)
            original_y = int((j - new_center_w) * math.sin(-theta) + (i - new_center_h) * math.cos(-theta) + center_h)

            if 0 <= original_x < image_w and 0 <= original_y < image_h:
                rotated_image[i,j] = image[original_y, original_x]

    return rotated_image

def edge_detection(image):
    sigma = 1.5
    filter_size = 5
    pad = 2
    gaus_filter = generate_gaussian(sigma, filter_size, filter_size)
    smooth_image = apply_filter(image, gaus_filter, pad, pad)

    enhance_image = hist_eq(smooth_image)

    threshold_value = 40
    threshold_image = numpy.where(enhance_image > threshold_value, 255, 0).astype(numpy.uint8)

    laplac_filter = numpy.array([[0, -1, 0], [-1, 4, -1], [0, -1, 0]])
    edge_image = apply_filter(threshold_image, laplac_filter, 1, 0)

    return edge_image