-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtransform.py
131 lines (106 loc) · 3.81 KB
/
transform.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
import numpy as np
from math import *
from numpy import linalg as LA
from scipy.spatial.transform import Rotation as R
#from quaternionop import quaternion_multiply, quaternion_divide
import cv2
'''
def slerp(quat1, quat2, alpha):
q_r=quaternion_divide(quat2,quat1)
w_r=q_r[3]
q_r= np.asarray(q_r)
if w_r<0 :
q_r= - q_r
v_r=q_r[0:3]
theta_r = 2*atan(LA.norm(v_r)/w_r)
n_r = v_r/LA.norm(v_r)
theta_alpha = alpha * theta_r
theta_n_r = sin(theta_alpha/2)*n_r
q_alpha = [theta_n_r[0], theta_n_r[1], theta_n_r[2], cos(theta_alpha/2)]
slerp1 = quaternion_multiply(q_alpha,quat1)
return slerp1
def rotation(R_1,R_2,num_frames, frame_num):
r1=R.from_dcm(R_1)
q1=r1.as_quat()
r2=R.from_dcm(R_2)
q2=r2.as_quat()
q_i=slerp(q1,q2,frame_num/num_frames)
r = R.from_quat(q_i)
R_i=r.as_dcm()
return R_i
def translation(t_1,t_2,num_frames, frame_num):
t=(t_2-t_1)*frame_num/num_frames+t_1
return t
def intrinsic(f1,f2,num_frames, frame_num,c_x_2,c_y_2):
f_nor=(f2-f1)*frame_num/num_frames+f1
f_inv=(f1-f2)*(num_frames-frame_num-1)/num_frames+f2
K2_nor = np.asarray([[f_nor,0,c_x_2],[0,f_nor,c_y_2],[0,0,1]])
K2_inv = np.asarray([[f_inv,0,c_x_2],[0,f_inv,c_y_2],[0,0,1]])
return K2_nor,K2_inv
def transform_normal(R_1,R_incr,t_1,t_incr,x,y,K1,K2):
R_t_init_1=np.identity(4)
R_t_init_1[0:3,0:3]=R_1
R_t_init_1[0:3,3]=t_1
K_ext_1=np.identity(4)
K_ext_1[0:3,0:3]=K1
P_1= K_ext_1@R_t_init_1
R_t_init_2=np.identity(4)
R_t_init_2[0:3,0:3]=R_incr
R_t_init_2[0:3,3]=t_incr
K_ext_2=np.identity(4)
K_ext_2[0:3,0:3]=K2
P_2= K_ext_2@R_t_init_2
[email protected](P_1)
x1=np.asarray([x,y,1,0]) #we consider that the depth d = 0
x2=M@x1
return x2[0], x2[1]
'''
def transform_corners(x1,y1,x2,y2,num_frames, frame_num):
x=(x2-x1)*frame_num/num_frames+x1
y=(y2-y1)*frame_num/num_frames+y1
return x,y
def find_corners(array):
idx_max_x=0
idx_max_y=0
idx_min_x=0
idx_min_y=0
max_x=array[0][0]
max_y=array[0][1]
min_x=array[0][0]
min_y=array[0][1]
for i in range (array.shape[0]):
if array[i][0]>max_x:
idx_max_x=i
max_x=array[i][0]
if array[i][1]>max_y:
idx_max_y=i
max_y=array[i][1]
if array[i][0]<min_x:
idx_min_x=i
min_x=array[i][0]
if array[i][1]<min_y:
idx_min_y=i
min_y=array[i][1]
return [idx_max_x,idx_max_y,idx_min_x,idx_min_y]
def warp_tri(inpt, tri1, tri2,output):
height_1,width_1=inpt.shape[:2]
r1 = cv2.boundingRect(tri1)
r2 = cv2.boundingRect(tri2)
tri1Cropped=[]
tri2Cropped=[]
for i in range(0, 3):
tri1Cropped.append(((tri1[i][0] - r1[0]),(tri1[i][1] - r1[1])))
tri2Cropped.append(((tri2[i][0] - r2[0]),(tri2[i][1] - r2[1])))
img1Cropped = inpt[r1[1]:r1[1] + r1[3], r1[0]:r1[0] + r1[2]]
warpMat = cv2.getAffineTransform(np.float32(tri1Cropped), np.float32(tri2Cropped))
img2Cropped = cv2.warpAffine( img1Cropped, warpMat, (r2[2], r2[3]), None, flags=cv2.INTER_LINEAR, borderMode=cv2.BORDER_REFLECT_101 )
mask = np.zeros((r2[3], r2[2], 3), dtype = np.float32)
cv2.fillConvexPoly(mask, np.int32(tri2Cropped), (1.0, 1.0, 1.0), 16, 0)
img2Cropped = img2Cropped * mask
for j in range(r2[1],r2[1]+r2[3]):
for i in range(r2[0],r2[0]+r2[2]):
if 0<i <width_1:
if 0<j<height_1:
if mask[j-r2[1],i-r2[0]][0]==1.0 and mask[j-r2[1],i-r2[0]][1]==1.0 and mask[j-r2[1],i-r2[0]][2]==1.0 :
output[j, i] = img2Cropped[j-r2[1],i-r2[0]]
return output