FloodImageClassifier/Flood_depth_estimation.py at main · durjoy33/FloodImageClassifier · 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
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
import cv2
import numpy as np
import math
from matplotlib import pyplot as plt
import skimage.viewer
import imutils

BLUE = (255,0,0)
original_img = ("./Images/flood_181.jpg")

def find_skyline(x1, y1, x2, y2, img):
    k = 1.0 * (y2 - y1) / 1.0 * (x2 - x1)
    d = 1.0 * (y1*x2 - y2*x1) / 1.0 * (x2 - x1)
    return(k,d)
def vignette_filter(img, pixels_falloff = 0, types = 0):
    height, width = img.shape
    radius = max(width, height) / 2.0 * 0.95
    radius = min(width, height) / 2.0 * 0.95
    #pixels_falloff = 0.1
    row_ctr = height / 2
    col_ctr = width / 2
    max_img_rad = math.sqrt(row_ctr * row_ctr + col_ctr * col_ctr)
    res = img.copy()

    if types:
        trow = pixels_falloff
        lcol = pixels_falloff
        brow = img.shape[0] - pixels_falloff * 2
        rcol = img.shape[1] - pixels_falloff * 2
    for i in range(height):
        for j in range(width):
            dh = abs(i - row_ctr)
            dw = abs(j - col_ctr)
            if not types:
                dis = math.sqrt(dh * dh + dw * dw)
                if dis > radius:
                    if dis > radius + pixels_falloff:
                        res[i, j] = img [i, j] * (dis) / radius
                        # cv2.imshow('res2', res)
                    else:
                        sigma = (dis - radius) / pixels_falloff
                        res[i, j] = img [i, j] * (1 - sigma * sigma)
                        # cv2.imshow('res3', res)
                else:
                    pass
            else:
                print('not here')
                dis1 = min(abs(i - trow), abs(i - brow))
                dis2 = min(abs(j - lcol), abs(j - rcol))
                if i <= brow and i >= trow and j >= lcol and j <= rcol:
                    pass
                else:
                    sigma = (dis1 + dis2) * (dis1 + dis2) / (dis1 * dis1 + dis2 * dis2)
                    res[i, j] = img [i, j] * sigma
    return res

def affline_rotate(img, pts1, pts2):
    rows = img.shape[0]
    cols = img.shape[1]
    M = cv2.getAffineTransform(pts1, pts2)
    dst = cv2.warpAffine(img,M,(cols,rows))
    return dst

if __name__ == "__main__":
    img = cv2.imread(original_img, 0)
    img = cv2.resize(img, (416,416))
    viewer = skimage.viewer.ImageViewer(image=img)
    viewer.show()

    img2 = vignette_filter(img, 0.3)
    viewer = skimage.viewer.ImageViewer(image=img2)
    viewer.show()

    img = cv2.blur(img, (5, 5))
    img2 = cv2.blur(img2, (10,10))

    img = cv2.addWeighted(img, 0.80, img2, 0.20, 1)
    #cv2.imshow('after vignette_filter', img)

    img = cv2.blur(img, (15,15))
    clahe = cv2.createCLAHE(clipLimit = 2.00, tileGridSize = (11,11))
    img = clahe.apply(img)
    ret2, detected_edges = cv2.threshold(img, 9, 255, cv2.THRESH_BINARY+cv2.THRESH_OTSU)
    #cv2.imshow('de', detected_edges)

    edges = cv2.Canny(detected_edges, 0.2, 1.8, apertureSize = 3)
    #cv2.imshow('ee', edges)

    dst = cv2.bitwise_and(img, img, mask = edges)
    #cv2.imshow('dst', dst)

    minLineLength = 100
    maxLineGap = 10
    lines = cv2.HoughLinesP(edges, 10, np.pi/180, minLineLength, maxLineGap)
    a1, b1, a2, b2 = (0,0,0,0)
    dis = 0

    print(lines[0])
    for x1,y1,x2,y2 in lines[0]:
        if (x1-x2)*(x1-x2) + (y1-y2)*(y1-y2) > dis:
            a1,b1,a2,b2 = (x1,y1,x2,y2)
    (k, d) = find_skyline(a1, b1, a2, b2, img)

    print("line : ", k, d)

    for i in range(dst.shape[0]):
        for j in range(dst.shape[1]):
            pos = int(k * j + d)
            if i <= pos + 4:
                dst[i, j] = 0
    #viewer = skimage.viewer.ImageViewer(image=dst)
    #viewer.show()

    original = cv2.imread(original_img)
    original = cv2.resize(original, (416,416))
    for row in range(original.shape[0]):
        for bt in range(original.shape[1]):
            if dst[row, bt] == 0:
                pass
            else:
                original[row, bt] = BLUE

    srcy1 = img.shape[1]
    srcx1 = k * srcy1 + d
    srcx2 = img.shape[0]
    srcy2 = 0
    srcx3 = img.shape[0]
    srcy3 = img.shape[1]
    pts1 = np.float32([[int(srcy1), int(srcx1)], [srcy2, srcx2], [srcy3, srcx3]])
    pts2 = np.float32([[img.shape[1] * 0.9, 0], [0, img.shape[0] / 7], [img.shape[1] * 0.75, img.shape[0]]])

    img = cv2.imread(original_img, 0)
    img2 = cv2.imread(original_img)
    img = cv2.resize(img, (416,416))
    #img = affline_rotate(img,pts1, pts2)
    #detected_edges = affline_rotate(detected_edges, pts1, pts2)
    #result = affline_rotate(original, pts1,pts2)
    viewer = skimage.viewer.ImageViewer(image=img)
    viewer.show()
    viewer = skimage.viewer.ImageViewer(image=detected_edges)
    viewer.show()
    #viewer = skimage.viewer.ImageViewer(image=result)
    #viewer.show()


    contours = cv2.findContours(detected_edges.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    contours = imutils.grab_contours(contours)
    bottle_clone = img.copy()
    cv2.drawContours(bottle_clone, contours, -1, (255, 0, 0), 2)
    viewer = skimage.viewer.ImageViewer(image=bottle_clone)
    viewer.show()

    areas = [cv2.contourArea(contour) for contour in contours]
    (contours, areas) = zip(*sorted(zip(contours, areas), key=lambda a:a[1]))
    print("Masked region area is :" + str(areas[-1]))
    image_area = img.shape[0] * img.shape[1]
    print("Total image area" + str(image_area))
    area_ratio = (areas[-1] / image_area) * 100
    print("Th percentage of the image occupied by the masked portion is : " + str(area_ratio))
    # print contour with largest area
    bottle_clone = img.copy()
    cv2.drawContours(bottle_clone, [contours[-1]], -1, (255, 0, 0), 2)
    (x, y, w, h) = cv2.boundingRect(contours[-1])
    cv2.putText(bottle_clone, "Area : " + str(areas[-1]), (x + 45 , y + 360), cv2.FONT_HERSHEY_PLAIN, 3, (255, 0, 0), 2)
    viewer = skimage.viewer.ImageViewer(image=bottle_clone)
    viewer.show()

    bottle_clone = img.copy()
    bottle_clone_1 = img2.copy()
    (x, y, w, h) = cv2.boundingRect(contours[-1])
    aspectRatio = w / float(h)
    print(aspectRatio)
    if aspectRatio >= 1.8:
        cv2.rectangle(bottle_clone, (x, y), (x + w, y + h), (0, 255, 0), 2)
        cv2.putText(bottle_clone, "level 1", (x + 45, y + 60), cv2.FONT_HERSHEY_PLAIN, 3, (255, 255, 0), 2)
    elif ((aspectRatio >= 1.62) and (aspectRatio < 1.8)):
        cv2.rectangle(bottle_clone, (x, y), (x + w, y + h), (0, 255, 0), 2)
        cv2.putText(bottle_clone, "level 2", (x + 45, y + 60), cv2.FONT_HERSHEY_PLAIN, 3, (255, 255, 0), 2)
    elif ((aspectRatio >= 1.44) and (aspectRatio < 1.62)):
        cv2.rectangle(bottle_clone, (x, y), (x + w, y + h), (0, 255, 0), 2)
        cv2.putText(bottle_clone, "level 3", (x + 45, y + 60), cv2.FONT_HERSHEY_PLAIN, 3, (255, 255, 0), 2)
    elif ((aspectRatio >= 1.26) and (aspectRatio < 1.44)):
        cv2.rectangle(bottle_clone, (x, y), (x + w, y + h), (0, 255, 0), 2)
        cv2.putText(bottle_clone, "level 4", (x + 45, y + 60), cv2.FONT_HERSHEY_PLAIN, 3, (255, 255, 0), 2)
    elif ((aspectRatio >= 1.08) and (aspectRatio < 1.26)):
        cv2.rectangle(bottle_clone, (x, y), (x + w, y + h), (0, 255, 0), 2)
        cv2.putText(bottle_clone, "level 5", (x + 45, y + 60), cv2.FONT_HERSHEY_PLAIN, 3, (255, 255, 0), 2)
    elif ((aspectRatio >= 0.90) and (aspectRatio < 1.08)):
        cv2.rectangle(bottle_clone, (x, y), (x + w, y + h), (0, 255, 0), 2)
        cv2.putText(bottle_clone, "level 6", (x + 45, y + 60), cv2.FONT_HERSHEY_PLAIN, 3, (255, 0, 0), 2)
    elif ((aspectRatio >= 0.72) and (aspectRatio < 0.90)):
        cv2.rectangle(bottle_clone, (x, y), (x + w, y + h), (0, 255, 0), 2)
        cv2.putText(bottle_clone, "level 7", (x + 45, y + 60), cv2.FONT_HERSHEY_PLAIN, 3, (255, 255, 0), 2)
    elif ((aspectRatio >= 0.54) and (aspectRatio < 0.72)):
        cv2.rectangle(bottle_clone, (x, y), (x + w, y + h), (0, 255, 0), 2)
        cv2.putText(bottle_clone, "level 3", (x + 45, y + 360), cv2.FONT_HERSHEY_PLAIN, 3, (255, 255, 0), 2)
    elif ((aspectRatio >= 0.36) and (aspectRatio < 0.54)):
        cv2.rectangle(bottle_clone, (x, y), (x + w, y + h), (0, 255, 0), 2)
        cv2.putText(bottle_clone, "level 9", (x + 45, y + 60), cv2.FONT_HERSHEY_PLAIN, 3, (255, 255, 0), 2)
    elif ((aspectRatio >= 0.18) and (aspectRatio < 0.36)):
        cv2.rectangle(bottle_clone, (x, y), (x + w, y + h), (0, 255, 0), 2)
        cv2.putText(bottle_clone, "level 10", (x + 45, y + 60), cv2.FONT_HERSHEY_PLAIN, 3, (255, 255, 0), 2)
    else:
        cv2.rectangle(bottle_clone, (x, y), (x + w, y + h), (0, 0, 255), 2)
        cv2.putText(bottle_clone, "Low", (x + 45, y + 60), cv2.FONT_HERSHEY_PLAIN, 3, (255, 0, 255), 2)
    image=cv2.cvtColor(bottle_clone,cv2.COLOR_GRAY2RGB)
    viewer = skimage.viewer.ImageViewer(image=image)
    viewer.show()