Skip to content

module utils/image_processing

image_processing

Functions

crop_region_with_margin 

crop_region_with_margin(image: np.ndarray, coordinates: tuple[int, int, int, int], margin_ratio: float = 0.2) -> np.ndarray

Crop a sub-region from an image with an additional margin.

Parameters:

Name Type Description Default
image np.ndarray

The input image (H, W, C) or (H, W).

 required
coordinates tuple[int, int, int, int]

The bounding box defined as (x1, y1, x2, y2).

 required
margin_ratio float

Ratio to determine the extra margin; default is 0.2.

 0.2

Returns:

Type Description
np.ndarray

The cropped image region including the margin.
Source code in color_correction/utils/image_processing.py 
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
def crop_region_with_margin(
    image: np.ndarray,
    coordinates: tuple[int, int, int, int],
    margin_ratio: float = 0.2,
) -> np.ndarray:
    """
    Crop a sub-region from an image with an additional margin.

    Parameters
    ----------
    image : np.ndarray
        The input image (H, W, C) or (H, W).
    coordinates : tuple[int, int, int, int]
        The bounding box defined as (x1, y1, x2, y2).
    margin_ratio : float, optional
        Ratio to determine the extra margin; default is 0.2.

    Returns
    -------
    np.ndarray
        The cropped image region including the margin.
    """
    y1, y2 = coordinates[1], coordinates[3]
    x1, x2 = coordinates[0], coordinates[2]

    height = y2 - y1
    margin_y = height * margin_ratio
    width = x2 - x1
    margin_x = width * margin_ratio

    crop_y1 = int(y1 + margin_y)
    crop_y2 = int(y2 - margin_y)
    crop_x1 = int(x1 + margin_x)
    crop_x2 = int(x2 - margin_x)

    return image[crop_y1:crop_y2, crop_x1:crop_x2]

calc_mean_color_patch 

calc_mean_color_patch(img: np.ndarray) -> np.ndarray

Compute the mean color of an image patch across spatial dimensions.

Parameters:

Name Type Description Default
img np.ndarray

The input image patch with shape (H, W, C).

 required

Returns:

Type Description
np.ndarray

Array containing the mean color for each channel (dtype uint8).
Source code in color_correction/utils/image_processing.py 
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
def calc_mean_color_patch(img: np.ndarray) -> np.ndarray:
    """
    Compute the mean color of an image patch across spatial dimensions.

    Parameters
    ----------
    img : np.ndarray
        The input image patch with shape (H, W, C).

    Returns
    -------
    np.ndarray
        Array containing the mean color for each channel (dtype uint8).
    """
    return np.mean(img, axis=(0, 1)).astype(np.uint8)

calc_color_diff 

calc_color_diff(image1: ImageBGR, image2: ImageBGR) -> dict[str, float]

Calculate color difference metrics between two images using CIE 2000.

Parameters:

Name Type Description Default
image1 ImageBGR

First input image in BGR format.

 required
image2 ImageBGR

Second input image in BGR format.

 required

Returns:

Type Description
dict[str, float]

Dictionary with keys 'min', 'max', 'mean', and 'std' for the color difference.
Source code in color_correction/utils/image_processing.py 
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
def calc_color_diff(
    image1: ImageBGR,
    image2: ImageBGR,
) -> dict[str, float]:
    """
    Calculate color difference metrics between two images using CIE 2000.

    Parameters
    ----------
    image1 : ImageBGR
        First input image in BGR format.
    image2 : ImageBGR
        Second input image in BGR format.

    Returns
    -------
    dict[str, float]
        Dictionary with keys 'min', 'max', 'mean', and 'std' for the color difference.
    """
    rgb1 = cv2.cvtColor(image1, cv2.COLOR_BGR2RGB)
    rgb2 = cv2.cvtColor(image2, cv2.COLOR_BGR2RGB)

    lab1 = cl.XYZ_to_Lab(cl.sRGB_to_XYZ(rgb1 / 255))
    lab2 = cl.XYZ_to_Lab(cl.sRGB_to_XYZ(rgb2 / 255))

    delta_e = cl.difference.delta_E(lab1, lab2, method="CIE 2000")

    return {
        "min": round(float(np.min(delta_e)), 4),
        "max": round(float(np.max(delta_e)), 4),
        "mean": round(float(np.mean(delta_e)), 4),
        "std": round(float(np.std(delta_e)), 4),
    }

numpy_array_to_base64 

numpy_array_to_base64(arr: np.ndarray, convert_bgr_to_rgb: bool = True) -> str

Convert a numpy image array into a base64-encoded PNG string.

Parameters:

Name Type Description Default
arr np.ndarray

Input image array.

 required
convert_bgr_to_rgb bool

Whether to convert BGR to RGB before encoding; default is True.

 True

Returns:

Type Description
str

Base64-encoded image string prefixed with the appropriate data URI.
Source code in color_correction/utils/image_processing.py 
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
def numpy_array_to_base64(
    arr: np.ndarray,
    convert_bgr_to_rgb: bool = True,
) -> str:
    """
    Convert a numpy image array into a base64-encoded PNG string.

    Parameters
    ----------
    arr : np.ndarray
        Input image array.
    convert_bgr_to_rgb : bool, optional
        Whether to convert BGR to RGB before encoding; default is True.

    Returns
    -------
    str
        Base64-encoded image string prefixed with the appropriate data URI.
    """
    if arr is None:
        return ""

    if convert_bgr_to_rgb:
        arr = cv2.cvtColor(arr, cv2.COLOR_BGR2RGB)
    img = Image.fromarray(arr)
    buffer = io.BytesIO()
    img.save(buffer, format="PNG")
    img_str = base64.b64encode(buffer.getvalue()).decode()
    return f"data:image/png;base64,{img_str}"