Montage-Generator/utils.py at main · 2513502304/Montage-Generator · GitHub

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
from skimage.color import deltaE_cie76, deltaE_ciede2000, deltaE_ciede94, deltaE_cmc
from typing import Sequence, MutableSequence
import numpy as np
import cv2 as cv
import os


def deltaE_manhattan(image1: Sequence[Sequence[int]], image2: Sequence[Sequence[int]], channel_axis: int = -1, weight: Sequence[int] = (1, 1, 1)) -> np.ndarray:
    """
    附带权重的两点之间的曼哈顿距离

    Args:
        image1 (Sequence[Sequence[int]]): image1
        image2 (Sequence[Sequence[int]]): image2
        channel_axis (int, optional): 参与运算的轴. Defaults to -1.
        weight (Sequence[int], optional): 图像通道权重. Defaults to (1, 1, 1).

    Returns:
        np.ndarray: 附带权重的两点之间的曼哈顿距离

    References:
        - https://wikimedia.org/api/rest_v1/media/math/render/svg/766971fac976a11f71166fb485df533072c886fb
    """
    c11, c12, c13 = np.moveaxis(image1.astype(np.float32), source=channel_axis, destination=0)[:3]
    c21, c22, c23 = np.moveaxis(image2.astype(np.float32), source=channel_axis, destination=0)[:3]
    c1_w, c2_w, c3_w = weight
    return c1_w * np.abs(c21 - c11) + c2_w * np.abs(c22 - c12) + c3_w * np.abs(c23 - c13)  # 最终结果与通道排列顺序无关


def deltaE_euclidean(image1: Sequence[Sequence[int]], image2: Sequence[Sequence[int]], channel_axis: int = -1, weight: Sequence[int] = (1, 1, 1)) -> np.ndarray:
    """
    附带权重的两点之间的欧几里得距离

    Args:
        image1 (Sequence[Sequence[int]]): image1
        image2 (Sequence[Sequence[int]]): image2
        channel_axis (int, optional): 参与运算的轴. Defaults to -1.
        weight (Sequence[int], optional): 图像通道权重. Defaults to (1, 1, 1).

    Returns:
        np.ndarray: 附带权重的两点之间的欧几里得距离

    References:
        - https://wikimedia.org/api/rest_v1/media/math/render/svg/766971fac976a11f71166fb485df533072c886fb
    """
    c11, c12, c13 = np.moveaxis(image1.astype(np.float32), source=channel_axis, destination=0)[:3]
    c21, c22, c23 = np.moveaxis(image2.astype(np.float32), source=channel_axis, destination=0)[:3]
    c1_w, c2_w, c3_w = weight
    return np.sqrt(c1_w * (c21 - c11)**2 + c2_w * (c22 - c12)**2 + c3_w * (c23 - c13)**2)  # 最终结果与通道排列顺序无关


def deltaE_approximation_bgr(image1: Sequence[Sequence[int]], image2: Sequence[Sequence[int]], channel_axis: int = -1) -> np.ndarray:
    """
    一种低成本近似方法
    这个公式的结果非常接近 L*u*v*（具有修改后的亮度曲线），更重要的是，它是一种更稳定的算法：它不存在一个颜色范围，在这个范围内会突然给出远离最优结果的结果

    Args:
        image1 (Sequence[Sequence[int]]): image1
        image2 (Sequence[Sequence[int]]): image2
        channel_axis (int, optional): 参与运算的轴. Defaults to -1.

    Returns:
        np.ndarray: RGB 色彩空间中两点之间的低成本近似距离

    Reference:
        - https://wikimedia.org/api/rest_v1/media/math/render/svg/95ee06baaa28944c5b1e06876439d1b579cf03c9
        - https://web.archive.org/web/20210327221240/https://www.compuphase.com/cmetric.htm
    """
    b1, g1, r1 = np.moveaxis(image1.astype(np.float32), source=channel_axis, destination=0)[:3]
    b2, g2, r2 = np.moveaxis(image2.astype(np.float32), source=channel_axis, destination=0)[:3]
    r_mean = (r1 + r2) / 2
    delat_r = r2 - r1
    delat_g = g2 - g1
    delta_b = b2 - b1
    return np.sqrt((2 + r_mean / 256) * delat_r**2 + 4 * delat_g**2 + (2 + (255 - r_mean) / 256) * delta_b**2)  # 最终结果与通道排列顺序有关


def deltaE_approximation_rgb(rgb1: Sequence[Sequence[int]], rgb2: Sequence[Sequence[int]], channel_axis: int = -1) -> np.ndarray:
    """
    一种低成本近似方法
    这个公式的结果非常接近 L*u*v*（具有修改后的亮度曲线），更重要的是，它是一种更稳定的算法：它不存在一个颜色范围，在这个范围内会突然给出远离最优结果的结果

    Args:
        rgb1 (Sequence[Sequence[int]]): rgb1
        rgb2 (Sequence[Sequence[int]]): rgb2
        channel_axis (int, optional): 参与运算的轴. Defaults to -1.

    Returns:
        np.ndarray: RGB 色彩空间中两点之间的低成本近似距离

    Reference:
        - https://wikimedia.org/api/rest_v1/media/math/render/svg/95ee06baaa28944c5b1e06876439d1b579cf03c9
        - https://web.archive.org/web/20210327221240/https://www.compuphase.com/cmetric.htm
    """
    r1, g1, b1 = np.moveaxis(rgb1.astype(np.float32), source=channel_axis, destination=0)[:3]
    r2, g2, b2 = np.moveaxis(rgb2.astype(np.float32), source=channel_axis, destination=0)[:3]
    r_mean = (r1 + r2) / 2
    delat_r = r2 - r1
    delat_g = g2 - g1
    delta_b = b2 - b1
    return np.sqrt((2 + r_mean / 256) * delat_r**2 + 4 * delat_g**2 + (2 + (255 - r_mean) / 256) * delta_b**2)  # 最终结果与通道排列顺序有关


def safe_imread(file: str) -> np.ndarray:
    """
    支持非 ASCII 编码路径的图像读取

    Args:
        file (str): 要读取的图像路径

    Returns:
        np.ndarray: 读取的 BGR 图像
    """
    # 以 uint8 数据类型，读取并解析二进制图像文件中的数据
    buffer = np.fromfile(file, dtype=np.uint8)
    # 解码数据为图像
    image = cv.imdecode(buffer, cv.IMREAD_COLOR)
    assert image is not None, f"Failed to load image: {file}"
    return image


def safe_imwrite(file: str, image: np.ndarray) -> True:
    """
    支持非 ASCII 编码路径的图像写入

    Args:
        file (str): 要写入的图像路径
        image (np.ndarray): 要写入的图像

    Returns:
        True: 是否写入成功
    """
    # 获取图像扩展名
    root, ext = os.path.splitext(file)
    # 将图像编码为指定文件格式的二进制数据
    ret, buffer = cv.imencode(ext, image)
    # 编码成功
    if ret:
        # 写入数据
        buffer.tofile(file)
        return True
    # 编码失败
    else:
        return False