-
Notifications
You must be signed in to change notification settings - Fork 0
/
State.py
84 lines (76 loc) · 2.68 KB
/
State.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
import numpy as np
import math
def get_m(act, arg):
[i, j, k] = arg
type = act[i, j, k]
a = j
b = k
# right
if type == 1:
a = j + 1
b = k
# left
if type == 2:
a = j - 1
b = k
# up
if type == 3:
a = j
b = k + 1
# down
if type == 4:
a = j
b = k - 1
# # rotate
# if type == 5:
# print('rotate')
# print(j, k)
# print(a, b)
# a = int(j * math.cos(1 / 180 * math.pi) + k * math.sin(1 / 180 * math.pi))
# b = int(k * math.cos(1 / 180 * math.pi) - j * math.sin(1 / 180 * math.pi))
# # -rotate
# if type == 6:
# a = int(j * math.cos(-1 / 180 * math.pi) + k * math.sin(-1 / 180 * math.pi))
# b = int(k * math.cos(-1 / 180 * math.pi) - j * math.sin(-1 / 180 * math.pi))
return a, b
class State():
def __init__(self):
self.warp_image = None
self.fixed_image = None
def reset(self, w, f):
self.warp_image = w
self.fixed_image = f
size = self.warp_image.shape
prev_state = np.zeros((size[0],64,size[2],size[3]),dtype=np.float32)
self.tensor = np.concatenate((self.warp_image - self.fixed_image, prev_state), axis=1)
# def set(self, m):
# self.warp_image = m
# # self.fixed = f
# self.tensor[:,:self.warp_image.shape[1],:,:] = self.warp_image
def step(self, act, inner_state):
tmp_img = np.zeros(self.warp_image.shape, dtype=np.float32)
b, c, h, w = self.warp_image.shape
for i in range(0, b):
# point
for j in range(0, h):
for k in range(0, w):
# if self.warp_image[i, 0, j, k] > 0:
a, b = get_m(act, [i, j, k])
a = min(a, w-1)
a = max(0, a)
b = min(b, w-1)
b= max(0, b)
if act[i, j, k] == 5:
tmp_img[i, 0, a, b] = self.warp_image[i, 0, j, k] + 1.0/255
if act[i, j, k] == 6:
tmp_img[i, 0, a, b] = self.warp_image[i, 0, j, k] - 1.0/255
if act[i, j, k] == 7:
tmp_img[i, 0, a, b] = self.warp_image[i, 0, j, k] + 10.0/255
if act[i, j, k] == 8:
tmp_img[i, 0, a, b] = self.warp_image[i, 0, j, k] - 10.0/255
else:
tmp_img[i, 0, a, b] = self.warp_image[i, 0, j, k]
# todo interpo
self.warp_image = tmp_img
self.tensor[:,:self.warp_image.shape[1],:,:] = self.warp_image - self.fixed_image
self.tensor[:,-64:,:,:] = inner_state