-
Notifications
You must be signed in to change notification settings - Fork 0
/
initialmatch.py
171 lines (157 loc) · 6.21 KB
/
initialmatch.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
import logging
from math import cos, pi
import numpy as np
from numpy.linalg.linalg import norm
import utils
from classes import Patch
from numpy import dot
import optim
import cv2 as cv
def run(images, alpha1, alpha2, omega, sigma, gamma, beta, filename, isDisplay) :
print("==========================================================", flush=True)
print(" INITIAL MATCHING ", flush=True)
print("==========================================================", flush=True)
# P <- empty
patches = []
# For each image I with optical center O(I)
for I in images :
# For each feature f detected in I
f_num = 1
for f in I.feats :
# F <- {Features satisfying the epipolar constraint}
F = computeF(I, images, f, omega, isDisplay)
# Sort F in an increasing order of distance from O(I)
sortF(F, f)
# For each feature f' in F
fp_num = 1
for fprime in F :
logging.info(f'IMAGE : {I.id+1:02d}/{len(images):02d}')
logging.info(f'FEAT : {f_num:02d}/{len(I.feats):02d}')
logging.info(f'FEAT\' : {fp_num:02d}/{len(F):02d}')
fp_num += 1
# Initialize c(p), n(p) and R(p)
p = computePatch(f, fprime, I)
# Initialize V(p) and V*(p)
Vp = computeVp(images, p, I, sigma)
VpStar = computeVpStar(Vp, p, alpha1, I)
if len(VpStar) < gamma :
logging.info("STATUS : FAILED")
logging.info("------------------------------------------------")
continue
# Refine c(p) and n(p)
new_p = refinePatch(p, VpStar, I)
# Update V(p) and V*(p)
Vp = computeVp(images, new_p, I, sigma)
VpStar = computeVpStar(Vp, new_p, alpha2, I)
# If |V*(p)| < gamma
if len(VpStar) < gamma :
# Fail
logging.info("STATUS : FAILED")
logging.info("------------------------------------------------")
continue
# Add p to P
patches.append(new_p)
# Add p to the corresponding Qj(x, y) and Qj*(x, y)
# Remove features from the cells where p was stored
registerPatch(new_p, Vp, VpStar, beta, f, fprime, False)
logging.info("STATUS : SUCCESS")
logging.info("------------------------------------------------")
# Exit innermost for loop
break
f_num += 1
utils.savePatches(patches, filename)
return patches
def computeF(I, images, f, omega, isDisplay) :
F = []
ref = cv.imread(I.name)
coord = np.array([f.x, f.y, 1])
if isDisplay :
cv.circle(ref, (coord[0], coord[1]), 4, (0, 255, 0), -1)
for image in images :
if I.id == image.id :
continue
else :
fmat = utils.fundamentalMatrix(I, image)
epiline = fmat @ coord
for feat in image.feats :
dist = utils.distance(feat, epiline)
if isDisplay :
img = image.displayFeatureMap()
epiline_x = (int(-epiline[2] / epiline[0]), 0)
epiline_y = (int((-epiline[2] - (epiline[1]*ref.shape[0])) / epiline[0]), ref.shape[0])
cv.line(img, epiline_x, epiline_y, (255, 0, 0), 1)
cv.circle(img, (feat.x, feat.y), 3, (0, 255, 0), -1)
cv.imshow(f'Ref : {I.id}', ref)
cv.imshow(f'Img : {image.id}, Dist : {dist}', img)
cv.waitKey(0)
cv.destroyAllWindows()
if dist <= omega:
F.append(feat)
return F
def sortF(F, f) :
for feat in F :
pt = utils.triangulate(f, feat, f.image.pmat, feat.image.pmat)
res = f.image.pmat @ pt
vec = pt - f.image.center
depth = norm(vec)
feat.depth = depth
utils.insertionSort(F)
def computePatch(f, fprime, ref) :
center = utils.triangulate(f, fprime, ref.pmat, fprime.image.pmat)
normal = ref.center - center
normal /= norm(normal)
patch = Patch(center, normal, ref)
return patch
def computeVp(images, patch, ref, sigma) :
Vp = []
Vp.append(ref)
for image in images :
if ref.id == image.id :
continue
else :
angle = (dot(patch.normal, (image.center - patch.center))) / (norm(image.center - patch.center))
if angle < cos(sigma * pi / 180) :
continue
else :
Vp.append(image)
return Vp
def computeVpStar(Vp, p, alpha, ref) :
VpStar = []
VpStar.append(ref)
for image in Vp :
if ref.id == image.id :
continue
else :
h = 1 - optim.computeDiscrepancy(ref, image, p)
if h < alpha :
VpStar.append(image)
return VpStar
def refinePatch(patch, VpStar, ref) :
refinedPatch = optim.run(patch, ref, VpStar)
return refinedPatch
def registerPatch(patch, Vp, VpStar, beta, f, fprime, isDisplay) :
for image in Vp :
pmat = image.pmat
pt = pmat @ patch.center
pt /= pt[2]
x = int(pt[0]/beta)
y = int(pt[1]/beta)
image.cells[y][x].q.append(patch)
isQStar = 0
if utils.getImage(image.id, VpStar) :
isQStar = 1
image.cells[y][x].qStar.append(patch)
patch.VpStar.append(image)
utils.removeFeatures(image, image.cells[y][x])
cell = np.array([image.id, [x, y], isQStar, 0])
patch.cells.append(cell)
patch.Vp.append(image)
if isDisplay :
ref = cv.imread(patch.ref.name)
cv.circle(ref, (int(f.x), int(f.y)), 3, (0, 255, 0), -1)
img = image.displayFeatureMap()
cv.circle(img, (int(pt[0]), int(pt[1])), 3, (0, 255, 0), -1)
cv.imshow(f'Ref : {patch.ref.id}', ref)
cv.imshow(f'Img : {image.id}', img)
cv.waitKey(0)
cv.destroyAllWindows()