-
Notifications
You must be signed in to change notification settings - Fork 0
/
fundus_prep.py
161 lines (134 loc) · 5.27 KB
/
fundus_prep.py
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
import numpy as np
import os
import cv2
def imread(file_path, c=None):
if c is None:
im = cv2.imread(file_path)
else:
im = cv2.imread(file_path, c)
if im is None:
raise 'Can not read image'
if im.ndim == 3 and im.shape[2] == 3:
im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB)
return im
def imwrite(file_path, image):
if image.ndim == 3 and image.shape[2] == 3:
image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
cv2.imwrite(file_path, image)
def fold_dir(folder):
if not os.path.exists(folder):
os.makedirs(folder)
return folder
def get_mask_BZ(img):
if img.ndim==3:
gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
else:
gray_img = img
threhold = np.mean(gray_img)/3-5
_, mask = cv2.threshold(gray_img, max(0,threhold), 1, cv2.THRESH_BINARY)
nn_mask = np.zeros((mask.shape[0]+2,mask.shape[1]+2),np.uint8)
new_mask = (1-mask).astype(np.uint8)
# cv::floodFill(Temp, Point(0, 0), Scalar(255));
# _,new_mask,_,_ = cv2.floodFill(new_mask, nn_mask, [(0, 0),(0,new_mask.shape[0])], (0), cv2.FLOODFILL_MASK_ONLY)
_,new_mask,_,_ = cv2.floodFill(new_mask, nn_mask, (0,0), (0), cv2.FLOODFILL_MASK_ONLY)
_,new_mask,_,_ = cv2.floodFill(new_mask, nn_mask, (new_mask.shape[1]-1,new_mask.shape[0]-1), (0), cv2.FLOODFILL_MASK_ONLY)
mask = mask + new_mask
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (20, 20))
mask = cv2.erode(mask, kernel)
mask = cv2.dilate(mask, kernel)
return mask
def _get_center_by_edge(mask):
center=[0,0]
x=mask.sum(axis=1)
center[0]=np.where(x>x.max()*0.95)[0].mean()
x=mask.sum(axis=0)
center[1]=np.where(x>x.max()*0.95)[0].mean()
return center
def _get_radius_by_mask_center(mask,center):
mask=mask.astype(np.uint8)
ksize=max(mask.shape[1]//400*2+1,3)
kernel=cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(ksize,ksize))
mask=cv2.morphologyEx(mask, cv2.MORPH_GRADIENT, kernel)
# radius=
index=np.where(mask>0)
d_int=np.sqrt((index[0]-center[0])**2+(index[1]-center[1])**2)
b_count=np.bincount(np.ceil(d_int).astype(np.int))
radius=np.where(b_count>b_count.max()*0.995)[0].max()
return radius
def _get_circle_by_center_bbox(shape,center,bbox,radius):
center_mask=np.zeros(shape=shape).astype('uint8')
tmp_mask=np.zeros(shape=bbox[2:4])
center_tmp=(int(center[0]),int(center[1]))
center_mask=cv2.circle(center_mask,center_tmp[::-1],int(radius),(1),-1)
# center_mask[bbox[0]:bbox[0]+bbox[2],bbox[1]:bbox[1]+bbox[3]]=tmp_mask
# center_mask[bbox[0]:min(bbox[0]+bbox[2],center_mask.shape[0]),bbox[1]:min(bbox[1]+bbox[3],center_mask.shape[1])]=tmp_mask
return center_mask
def get_mask(img):
if img.ndim ==3:
#raise 'image dim is not 3'
g_img=cv2.cvtColor(img,cv2.COLOR_RGB2GRAY)
elif img.ndim == 2:
g_img =img.copy()
else:
raise 'image dim is not 1 or 3'
h,w = g_img.shape
shape=g_img.shape[0:2]
g_img = cv2.resize(g_img,(0,0),fx = 0.5,fy = 0.5)
tg_img=cv2.normalize(g_img, None, 0, 255, cv2.NORM_MINMAX)
tmp_mask=get_mask_BZ(tg_img)
center=_get_center_by_edge(tmp_mask)
#bbox=_get_bbox_by_mask(tmp_mask)
radius=_get_radius_by_mask_center(tmp_mask,center)
#resize back
center = [center[0]*2,center[1]*2]
radius = int(radius*2)
s_h = max(0,int(center[0] - radius))
s_w = max(0, int(center[1] - radius))
bbox = (s_h, s_w, min(h-s_h,2 * radius), min(w-s_w,2 * radius))
tmp_mask=_get_circle_by_center_bbox(shape,center,bbox,radius)
return tmp_mask,bbox,center,radius
def mask_image(img,mask):
img[mask<=0,...]=0
return img
def remove_back_area(img,bbox=None,border=None):
image=img
if border is None:
border=np.array((bbox[0],bbox[0]+bbox[2],bbox[1],bbox[1]+bbox[3],img.shape[0],img.shape[1]),dtype=np.int)
image=image[border[0]:border[1],border[2]:border[3],...]
return image,border
def supplemental_black_area(img,border=None):
image=img
if border is None:
h,v=img.shape[0:2]
max_l=max(h,v)
if image.ndim>2:
image=np.zeros(shape=[max_l,max_l,img.shape[2]],dtype=img.dtype)
else:
image=np.zeros(shape=[max_l,max_l],dtype=img.dtype)
border=(int(max_l/2-h/2),int(max_l/2-h/2)+h,int(max_l/2-v/2),int(max_l/2-v/2)+v,max_l)
else:
max_l=border[4]
if image.ndim>2:
image=np.zeros(shape=[max_l,max_l,img.shape[2]],dtype=img.dtype)
else:
image=np.zeros(shape=[max_l,max_l],dtype=img.dtype)
image[border[0]:border[1],border[2]:border[3],...]=img
return image,border
def process_without_gb(img):
# preprocess images
# img : origin image
# tar_height: height of tar image
# return:
# result_img: preprocessed image
# borders: remove border, supplement mask
# mask: mask for preprocessed image
borders = []
mask, bbox, center, radius = get_mask(img)
r_img = mask_image(img, mask)
r_img, r_border = remove_back_area(r_img,bbox=bbox)
mask, _ = remove_back_area(mask,border=r_border)
borders.append(r_border)
r_img,sup_border = supplemental_black_area(r_img)
mask,_ = supplemental_black_area(mask,border=sup_border)
borders.append(sup_border)
return r_img,borders,(mask*255).astype(np.uint8)