forked from TheJaeLal/LineFormer
-
Notifications
You must be signed in to change notification settings - Fork 0
/
line_utils.py
252 lines (202 loc) · 7.9 KB
/
line_utils.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
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
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
import numpy as np
import cv2
from matplotlib import pyplot as plt
import scipy
# For line interpolation:
# Comment this out if not using 'get_interp_points' function
from bresenham import bresenham
def hsv_to_bgr(h, s, v):
# Get RGB values
c = v * s
x = c * (1 - abs((h * 6) % 2 - 1))
m = v - c
if h < 1/6:
r, g, b = c, x, 0
elif h < 1/3:
r, g, b = x, c, 0
elif h < 0.5:
r, g, b = 0, c, x
elif h < 2/3:
r, g, b = 0, x, c
elif h < 5/6:
r, g, b = x, 0, c
else:
r, g, b = c, 0, x
# Scale RGB values to 0-255 range and convert to integers
r = int((r + m) * 255)
g = int((g + m) * 255)
b = int((b + m) * 255)
return (b, g, r)
def get_distinct_colors(n):
huePartition = 1.0 / (n + 1)
return (hsv_to_bgr(huePartition * value, 1.0, 1.0) for value in range(0, n))
def is_color(img):
if img.ndim <=2:
return False
# img.ndim >=3
if img.shape[-1] == 1:
return False
return True
def show_img(img, color='gray', is_bgr=False, title='', figsize=None, final_show=True):
"""Show image using plt."""
if figsize:
if not isinstance(figsize, (tuple, list)):
figsize = (figsize, figsize)
plt.figure(figsize=figsize)
if is_color(img) and is_bgr:
img = img[...,::-1]
params = {'cmap':color}
if img.dtype == np.uint8:
params.update({'vmin': 0, 'vmax': 255})
plt.imshow(img, **params)
plt.title(title)
if final_show:
plt.show()
return
def draw_xrange(img, xrange):
annot_img = img.copy()
im_h, im_w = annot_img.shape[:2]
annot_img = cv2.line(annot_img, (xrange[0], 0), (xrange[0], im_h), (0,0,255), thickness=1)
annot_img = cv2.line(annot_img, (xrange[1], 0), (xrange[1], im_h), (0,0,255), thickness=1)
return annot_img
def get_xrange(bin_line_mask):
"""
bin_line_mask: np.ndarray => black and white binary mask of line
black => background => 0
white => foregrond line pixel => 255
returns: (x_start, x_end) where x_start and x_end represent the starting and ending points
for the binary line segment
"""
# print(bin_line_mask.sum(axis=0))
# np.save("problem_mask.npy", bin_line_mask)
smooth_signal = scipy.signal.medfilt(bin_line_mask.sum(axis=0), kernel_size=5)
# print(smooth_signal.shape)
# print(smooth_signal)
# print(np.nonzero(smooth_signal))
x_range = np.nonzero(smooth_signal)
if len(x_range) and len(x_range[0]): # To handle cases with empty masks
x_range = x_range[0][[0, -1]]
else:
x_range = None
return x_range
def get_kp(line_img, interval=10, x_range=None, get_num_lines=False, get_center=True):
"""
line_img: np.ndarray => black and white binary mask of line
black => background => 0
white => foregrond line pixel => 255
interval: delta_x at which x,y points are sampled across the line_img
x_range: Range of x values, [xmin, xmax), within which pred points (x,y) are to be sampled
returns: a list [{'x': <x_val>, 'y': <y_val>}, ....] of line points found in the binary line_img
"""
im_h, im_w = line_img.shape[:2]
kps = []
# delta = 2
if x_range is None:
x_range = (0, im_w)
# track the number of vertical binary components found at every x => estimate num lines
num_comps = []
for x in range(x_range[0], x_range[1], interval):
# get the corresponding white pixel in this column
fg_y = []
fg_y_center = []
all_y_points = np.where(line_img[:, x] == 255)[0]
if all_y_points.size != 0:
fg_y.append(all_y_points[0])
y = all_y_points[0]
n_comps = 1
for idx in range(1, len(all_y_points)):
y_next = all_y_points[idx]
# print(y, y_next)
if abs(y_next - y) > 2:
n_comps += 1
# break found b/w y_next and y, separate components
if fg_y[-1] != y:
# handle the case where (first component itself is broken, i.e found break at idx=1)
fg_y_center.append(round(y + fg_y[-1])//2)
fg_y.append(y)
else:
fg_y_center.append(y)
fg_y.append(y_next)
y = y_next
# print(fg_y,'\n', fg_y_center, '\n')
# print('last_point', y, y_next)
if fg_y[-1] != y:
# add the last point
fg_y_center.append(round(y + fg_y[-1])//2)
fg_y.append(y)
else:
fg_y_center.append(y)
num_comps.append(n_comps)
if (fg_y or fg_y_center) and (n_comps==1):
if get_center:
kps.extend([{'x':float(x), 'y':y} for y in fg_y_center])
else:
kps.extend([{'x':float(x), 'y':y} for y in fg_y])
res = kps
if get_num_lines:
res = kps, int(np.percentile(num_comps, 85))
return res
def draw_edge(img, edge):
inter_points = get_interp_points(edge[0], edge[1])
# print(inter_points)
# print(len(inter_points), inter_points)
annot_img = draw_kps(img, array_to_points(inter_points), color=(255,0,0))
return annot_img
def draw_kps(img, kps, color=(0,255,0), classes=None, **draw_options):
if is_color(img):
annot_img = img.copy()
else:
annot_img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
if classes is None:
classes = [0]*len(kps)
color_map = {0: color}
else:
colors = list(get_distinct_colors(classes.max()+1))
color_map = dict(zip(range(classes.max()+1), colors))
# print(color_map)
for idx, kp in enumerate(kps):
options = dict(color=color_map[classes[idx]], markerType=cv2.MARKER_CROSS, markerSize=2, thickness=2, line_type=8)
options.update(draw_options)
annot_img = cv2.drawMarker(annot_img, (int(kp['x']), int(kp['y'])), **options)
return annot_img
def points_to_array(pred_ds):
res = []
for line in pred_ds:
line_arr = []
for pt in line:
line_arr.append([pt['x'], pt['y']])
res.append(line_arr)
return res
# res = line_utils.draw_lines(img, points_to_array(pred_ds))
def draw_lines(img, lines, classes=None):
if is_color(img):
annot_img = img.copy()
else:
annot_img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
if classes is None:
classes = list(range(len(lines)))
if len(classes):
colors = list(get_distinct_colors(max(classes)+1))
color_map = dict(zip(range(max(classes)+1), colors))
# print('color_map:', color_map)
for line_idx, line in enumerate(lines):
options = dict(color=color_map[classes[line_idx]], thickness=2)
drawing_lines = []
for pt_idx in range(len(line)-1):
drawing_lines.append([line[pt_idx], line[pt_idx+1]])
annot_img = cv2.polylines(annot_img, np.array(drawing_lines), isClosed=False, **options)
return annot_img
# Get the line points that would lie between ptA and ptB, according to the bresenham algorithm
def get_interp_points(ptA, ptB, thickness=1):
# x_interp = np.arange(ptA[0], ptB[0])
# y_interp = np.interp(x_interp, [ptA[0], ptB[0]], [ptA[1], ptB[1]]).round().astype(int)
points = []
delta_range = (-thickness//2, thickness//2)
for delta in range(delta_range[0], delta_range[1]+1):
points.extend(list(bresenham(ptA[0], ptA[1]+delta, ptB[0], ptB[1]+delta)))
inter_points = np.array(points)
# inter_points = np.stack([x_interp,y_interp], axis=-1)
return inter_points
def array_to_points(pts_arr):
pts = [{'x': pt[0], 'y': pt[1]} for pt in pts_arr]
return pts